**Registration for the conference is open yet. Go to icce.org.
**Don't forget to submit your copyright
**Signed letters of invitation can be created and printed in EDAS
Program for 2018 IEEE International Conference on Consumer Electronics (ICCE)
Friday, January 12
Friday, January 12 8:30 - 9:00
Friday, January 12 9:00 - 9:30
Friday, January 12 9:30 - 10:30
Keynote 1: Smart Healthcare,
Niraj K. Jha, Department of Electrical Engineering,
- 9:30 Smart Healthcare
- In this talk, we will explore smart healthcare from two angles: energy-efficient inference on WMSs and use of such sensors in various applications. Energy-efficient inference is made possible on sensor nodes by exploiting sparsity, which is characteristic of a signal that allows us to represent information efficiently. We will look at an approach that enables efficient representations based on sparsity to be utilized throughout a signal processing system, with the aim of reducing the energy and/or resources required for computation, communication, and storage. Such intelligent WMSs can be expected to be an important pillar of smart healthcare. We will then explore the use of WMSs in various applications: energy- and storage-efficient continuous health monitoring, medical diagnosis in the context of a health decision support system, stress detection/alleviation, etc. Finally, we will discuss some of the security issues associated with smart healthcare.
Friday, January 12 10:30 - 11:00
- 10:30 A QoE-aware Quality Selection Controller for HTTP Adaptive Streaming
- In this paper, we propose a quality of experience (QoE)-aware quality selection controller for HTTP adaptive streaming. The proposed method improves QoE by using two types of control: insertion of intermediate quality and suppression of quality switching. We implement the proposed method on dynamic adaptive streaming over HTTP (DASH) system and evaluate QoE. We show that the proposed method improves QoE by subjective evaluation using the scale of absolute category rating (ACR) method.
- 10:33 QoE-aware Quality Selection Method for Adaptive Video Streaming with Scalable Video Coding
- In this paper, we propose a quality of experience (QoE) aware quality-level selection method for adaptive video streaming with scalable video coding (SVC). The proposed method consists of the following two factors: segment groups and a buffer-aware layer selection algorithm. The proposed method uses these two factors to prevent features which degrade QoE performance. We implement the proposed algorithm on a SVC-DASH system and evaluate the QoE performance of the proposed method via subjective evaluations of multiple users. As a result, we show that the proposed method is effective to improve QoE performance.
- 10:37 Multi-Frame Super Resolution Using Frame Selection and Multiple Fusion for 250 Million Pixel Images
- In this paper, we originally propose an adaptive frame selection method and a multiple fusion scheme to enhance the quality of 250 million pixel images. This method is quite suitable for commercial use due to accuracy, simplicity and ease of implementation.
- 10:41 Low-Power Image Stitching Management for Reducing Power Consumption of UAVs for Disaster Management System
- Recently, unmanned aerial vehicles (UAVs) are used for disaster management system that monitors disasters (e.g., forest fire and landslide) of some areas (e.g., mountains) and responds to disasters. The UAVs of the disaster management system take images and sensor data (e.g., temperature and humidity data) of the areas where the disasters can occur, and then the disaster management system stitches the images of the areas to monitors the areas. This paper proposes the low-power image stitching management (LPISM) that can reduce power consumptions of the UAVs that take images for image stitching in the disaster management system. The disaster management system with the LPISM generates the least number of waypoints for the UAV to take images for image stitching. Therefore, the UAVs can reduce power consumption for taking images, and the UAV can increase the frequency of gathering sensor data for monitoring disasters in detail.
- 10:45 A Receiver Position Estimation Scheme in Wireless Power Transfer System
- The multiple magnetic resonance coils are effective at transmitting wireless power to a single receiver. However, the receiver's location is essential to the transmitter for power transmission efficiency. In this paper, we propose a position estimation scheme for magnetic multiple-input single-output (MISO) wireless power transfer (WPT) system. The MISO WPT system with the fixed frequency at 6.78MHz is applied in this work. We construct an offline training data set as the reference. Then, a KNN algorithm is employed to search the nearest instance with the current measurement. With such location information, the efficiency of wireless power transfer can be enhanced.
- 10:49 A New Auto-Brightness Control Technology for Transparent Displays
- We propose a new auto-brightness control technology for transparent displays based on a linear relation between the best luminance and an eye illuminance measured near eyes facing the display.
- 10:52 Hybrid Application Service with Companion Devices Based on T-UHDTV
- In this paper, we introduce the design and flow scheme for a Hybrid Application Service system, which can provide broadcasting service to the viewer's private device using parsed information via Primary Device receiver. Our proposed system is extended based on Dynamic Linkage Service which is a kind of hybrid broadcasting service.
- 10:56 New Calibration Method of Plenoptic Camera Using CCD Camera Model
- This paper presents a convenient method to estimate the internal parameters of a plenoptic camera by applying a CCD camera calibration method and some formulas to be analytically derived. A microlens based plenoptic camera is effectively same as a virtual camera array placed in regular intervals. For calibration, we employ a homogeneous intrinsic matrix based on the plenoptic camera model. Using various check board patterns, as in a CCD camera calibration, the disparities of pixels, the center positions of the virtual cameras, the baseline between the virtual cameras, and the distance to a point in the scene are extracted in the sub-aperture image domains. We formulate four equations for calibration of the plenoptic camera and then performs a nonlinear optimization technique to find a solution to these equations. To evaluate the estimation performance, we generate an artificial plenoptic image whose internal parameters are definite. Then estimation results are compared with the true parameters. In addition, the reprojection error by the estimated parameters is demonstrated.
Friday, January 12 11:00 - 12:00
- 11:00 Camera-recognizable Visual Cryptography for Printed Documents: Enhancing the User Interface
- Sensitive information in printed documents are susceptible to security risks. This paper proposes a system that creates an image-based cypher on printed documents and interprets image cyphers for users. Our evaluation results show that the proposed system satisfies important requirements such as security, usability, and performance.
- 11:20 Secure and Robust User Authentication Using Partial Fingerprint Matching
- this paper proposes a robust fingerprint-based authentication algorithm that ensures secure verification even with limited-sized partial fingerprints. It has become increasingly common that a number of recent consumer devices, such as smartphones, employ fingerprint sensors for user authentication. The sensors to be embedded are generally preferred to take up less space for better usability and product design, the sensing areas are therefore limited in size. To supplement the insufficient area, devices often store multiple acquisitions from a single finger in enrollment, later to verify at least any one of them successfully match an acquisition in authentication. Considering the low information entropy of a partial image, the security aspect of small area-based systems is a major concern. On the other hand, unpredictable variability due to finger rotation, grip positions, and skin deformation has a negative impact on biometric performance. This paper proposes a fingerprint matcher or verifier incorporating several efficient algorithms against these concerns on both the performance and security aspects. A method of "segmented area matching" brought an enhanced robustness to the variability, especially to finger rotation. And a method of "feature-weighted block correlation scoring" provided more detailed image discrimination, resulting in improved security. Experimental evaluations with extensive database of partial fingerprint images from more than 100 people, acquired by a small-sized capacitive sensor, demonstrated a significant improvement over the previously suggested algorithms in both aspects.
- 11:40 Strong Privacy Preserving Authentication Scheme for Unmanned Cars
- The vehicular ad hoc network (VANET) has several security issues. Identification requires accurate authentication, but privacy must be kept secretly. In this paper, we propose a strong privacy preserving authentication scheme to solve these problems based on vehicles to infrastructures (V2I) communications. We introduced public key infrastructure (PKI) and message accessing code (MAC) for encryption. We define adversary model in two types external and internal. We demonstrate how our methods are safe for these attackers.
- 11:00 Real-Time Human Activity-Based Energy Management System Using Model Predictive Control
- The challenges of energy efficiency and comfort management of intelligent homes and buildings are usually tackled with methods relying on historical data and a large number sensors. In this paper, we propose a real-time human activity-based energy management system (HAEMS), which tracks and processes human movement, and achieves device control based on real-time data with a significantly reduced number of sensors. A human activity detection algorithm and a model predictive control scheme are developed and implemented to optimally manage energy. A multi-objective optimization problem is formulated to minimize electricity cost and control temperature for thermal comfort. The HAEMS is deployed in a scaled-down laboratory setup and the performance is evaluated in an embedded system and hardware environment. Experimental results show that this system is able to optimize both electricity cost and thermal comfort.
- 11:15 A Pixel-based Complexity Model to Estimate Energy Consumption in Video Decoders
- The increasing use of HEVC video streams in diverse multimedia applications is driving the need for higher user control and management of energy consumption in battery-powered devices. This paper presents a contribution for the lack of adequate solutions by proposing a pixel-based complexity model that is capable of estimating the energy consumption of an arbitrary software-based HEVC decoder, running on different hardware platforms and devices. In the proposed model, the computational complexity is defined as a linear function of the number of pixels processed by the main decoding functions, using weighting coefficients which represent the average computational effort that each decoding function requires per pixel. The results shows that the cross-correlation of frame-based complexity estimation with energy consumption is greater than 0.86. The energy consumption of video decoding is estimated with the proposed model within an average deviation range of about 6.9%, for different test sequences.
- 11:30 Bandwidth-Aware DRAM Page Migration for Heterogeneous Mobile Memory Systems
- Bandwidth demands for mobile consumer electronics devices, including smartphones and tablets, have been continuously increasing in recent years. A heterogeneous memory system (HMS) integrating different types of mobile DRAM devices, such as LPDDR4 and WideIO2, is an attractive solution to scale DRAM bandwidth beyond the maximum bandwidth of a single device type. One of the key design challenges in an HMS is how to identify hot (cold) pages and migrate them to the faster (slower) devices at a low cost. This becomes more difficult as more devices like LPDDR4X adopt dynamic voltage and frequency scaling (DVFS) for DRAM I/O to reduce power consumption. At a low I/O voltage and frequency, a significant portion of DRAM bandwidth can be consumed by page migration traffic to slow down user applications. Thus, this paper proposes a novel page migration technique towards bandwidth-aware memory management (BAMM) for an HMS composed of both fast WideIO2 and slow LPDDR4 devices. BAMM periodically measures the bandwidth consumption for the LPDDR4 channel and adjusts both the maximum cap for migration bandwidth and DVFS level to keep the migration overhead low. Our evaluation using a detailed cycle-level simulator demonstrates that BAMM improves the energy efficiency of the memory system by 6.3% on average (and up to 11.6%) with minimal performance overhead over the baseline HMS with no dynamic page migration.
- 11:45 Non Intrusive Load Identification for Smart Energy Management Systems
- In an energy management (EM) perspective, an objective function considering the balance of energy and resources needs to be optimized. To reach the goal, in a residential scenario, a number of tasks (typically appliances) can be managed (e.g. shifted). This paper proposes an unobtrusive cascade classifier approach to autonomously identify the different appliances and their priorities (shiftable, non interruptible, continuous loads) in EM algorithms based on smart plugs measures.
- 11:00 Critical Areas Detection and Vehicle Speed Estimation System Towards Intersection-Related Driving Behavior Analysis
- A large number of serious traffic accidents occur at intersections due to the unsafe driving behaviors. In this paper, we propose a smartphone-based system to provide important information for driving behavior analysis at intersections. Our proposed system consists of two parts: (1) a deep convolutional neural network based model to detect traffic lights, crosswalks, and stop lines. (2) a Long Short-Term Memory (LSTM) neural network based model to estimate vehicle speed using accelerometer and gyroscope embedded in the smartphone. Important objects detection in traffic scenes and real time vehicle speed estimation are crucial for driving behavior analysis. We performed a thorough evaluation of our system, including analysis of the effectiveness of the proposed algorithm itself and the comparison with other methods as well. Our experiments exhibit the robustness of our system in various traffic scenarios.
- 11:20 On the Successful Deployment of Community Policing Services the TRILLION Project Case
- The evolution of policing towards the next generation, not only involves confronting new types of crime such as cybercrime, but also the active engagement of citizens in the process of creating a secure environment through the deployment of community policing practices. However, in order to fully exploit the potential of community policing, building of trust between citizens and Law Enforcement Officers is important. In this paper, the authors (all participating in the EU Research project on community policing TRILLION), discuss issues related to the best use of technology, while ensuring societal approval, in the context of deployment of a platform for community policing. Both requirements and corresponding work leading to the actual implementation of a fully operational platform are presented.
- 11:40 An Efficient Automatic Electrocardiogram Analysis Method Using Smartphones
- Long-term electrocardiogram (ECG) is an important diagnostic assistant approach in capturing the intermittent cardiac arrhythmias. The combination of miniaturized Holter and healthcare platforms enable people to have their cardiac arrhythmias monitored at home. The high computational burden created by the synchronized daily schedule of numerous users poses a severe challenge to the healthcare platform. Thus, shifting analysis tasks from healthcare platform to mobile computing devices are considered.However, long-term ECG data processing on smartphone/tablets can't meet the demands of real-time response due to the limited computing capability. In this paper, we developed a novel parallel automatic ECG analysis algorithm using the open computing language (OpenCL) framework. The experimental results show that, comparing to 7.57s of the sequential automatic ECG analysis algorithm, the executing time of the proposed parallel algorithm on 24-hour-long ECG data reduces to 1.45s, which achieves a speedup of 5.22x without reducing analysis accuracy. Furthermore, average power of the parallel algorithm is less than that of the sequential algorithm, resulting in saving about 61% of the battery energy. The problem of limited battery working hours for mobiles devices would be alleviated.
ST05-1: Architecture-Level Energy, Security, and Reliability Solutions for CE Digital Hardware (ESR) - Session 1
- 11:00 Multi-Phase Watermark for IP Core Protection
- Embedding a strong watermark is sufficient to prove IP core ownership during conflict resolution process. However, watermark embedded at lower levels of abstraction may incur design overhead and complexity. Further, watermark embedded at lower level of design does not help to protect a reusable IP core generated during architectural synthesis (at higher abstraction level). This paper presents a low overhead multi-phase watermark implanted during architectural synthesis that is more robust and tamper tolerant than existing single phase watermarks at higher abstraction level.
- 11:20 Reusable Intellectual Property Core Protection for Both Buyer and Seller
- This paper presents a methodology for IP core protection of CE devices from both buyer's and seller's perspective. In the presented methodology, buyer fingerprint is embedded along seller watermark during architectural synthesis phase of IP core design. The buyer fingerprint is inserted during scheduling phase while seller watermark is implanted during register allocation phase of architectural synthesis process. The presented approach provides a robust mechanisms of IP core protection for both buyer and seller at zero area overhead, 1.1 \% latency overhead and 0.95 \% design cost overhead compared to a similar approach (that provides only protection to IP seller).
- 11:40 RelBat: A Reliable Battery System Towards the Realization of Sustainable Electronics
- This paper presents a battery simulation flow and system to help design engineers to simulate battery arrays, demonstrate their characteristics, their modes of operation and the results of their implementation. As an example application, a reliable battery is presented which consists of battery cells, a cell switching circuit, a battery cell array manager and a system safety manager. This is useful for managing the existing energy infrastructure in an efficient and smart manner such that batteries last for a longer time. A RelBat system is built on a management system which performs its management activities by protecting the battery from operating outside its safe operating area, monitoring its state, calculating data, reporting the data and controlling its environment (modifying the temperature). Related work has focused on a battery model system which adapts to the current profile of the device, while this research focuses on the deployment of smarter batteries to self-manage lifetime.
- 11:00 Smart-Log: An Automated, Predictive Nutrition Monitoring System for Infants Through IoT
- Malnutrition is a condition where the body deprives of important nutrients required to maintain healthy tissues and organ function. Maintaining the right balance in the food intake is very important especially in infants where tremendous growth occurs. Unlinke adults, infants require someone's assistance in their food intake. In the present world, where most of the infants are being sent to Day-care, an automated food monitoring system helps in keeping track of their food intake. In this paper an automated food monitoring system with predictions to help a balanced meal is proposed. This sensor system consists of a piezo based sensor board which can help in analyzing the weight of each meal and a smart phone camera to obtain Nutrition Facts of the ingredients.
- 11:20 Smart-Walk: An Intelligent Physiological Monitoring System for Smart Families
- individuals. Understanding the lifestyle of family and kin can help in improving quality of their life. Wearables for activity tracking and wrist-worn devices have been an ever-growing Industrial sector. But with multiple functionalities embarked in a small device, the accuracy and power consumption can be compromised. This paper proposes a piezo-electric based accelerometer sensor design framework which helps in tracking the physical activities of family and friends. The accuracy of the human activity algorithm is analyzed by the various parameters analyzed through the sensor output. This proposed framework was validated using TIMSP432, Educational BoosterPack and \matlab and the learning parameters were modeled using WEKA. The feature based human activity monitoring algorithm gives 97.9\% efficiency in worst case.
- 11:40 IoT Based Indoor Location Detection System for Smart Home Environment
- Smart home environment is expected to meet the requirements, essentially for the aging population, to support the concept of "aging in place", to provide reliable care and to ensure safety and proper diagnosis by keeping track of daily living, medical condition of the resident and providing feedback to the caregiver. In order to meet these requirements, smart home of today should support a number of functionalities. One such functionality of a smart home environment is the location detection. In this work, we propose a voice based location detection system which can be integrated in a smart home environment. Our location detection system uses Amazon Echo as the voice interface and HC-SR04 ultrasonic sensor to detect location of specific patients. The proposed location detection will be suitable for large scale application where we may need to keep track of multiple patients. Moreover, the inclusion of voice enabled feature to this system will reduce the burden of learning curve of new technologies on family and caregivers thereby improving the quality of life.
Friday, January 12 12:00 - 12:30
Friday, January 12 12:30 - 13:30
- 12:30 2020 Life with 50 Billion Connected Devices
- It's a sunny March morning in 2023. 7:18 AM. You're buttering your bagel and gulping your coffee like always. You're looking forward to your commute - you'll do a conference call with your team and catch up on the news. You're thankful you don't have to actually drive, your car knows the way. As you get into your car you are presented with an urgent message. Your car has been immobilized and you need to pay 4 Bitcoin in ransom. You're not going anywhere right now. The world will have 50 billion connected devices by 2020. We've all heard this stat by now, but what does this really mean for individuals and society? What will be different? How fast will these shifts occur? Will we be ready? Learn from one of the foremost IoT thought leaders how a world of sensors, devices and machines everywhere, some we see, others we don't, sending vast quantities of data, will affect our daily lives, change our behaviors, and influence our thoughts about innovation, convenience, security and privacy. We'll examine a day in the life of a digital citizen in 2020 and identify the implications of a world where nearly everything is connected. 3 questions this session will answer: • What does a world of 50 billion connected devices look like? What are all these devices, what are they doing and why? • What are some of the major ways society and interpersonal relations will change in a world where nearly everything is connected? • This world is coming - are we ready? What are some of the implications of this onslaught of connectivity on issues like privacy and security?
Friday, January 12 13:30 - 14:30
- 13:30 Constructing Universal Secure and Private Data Networks Across Consumer Systems
- Software Constructed Push-Pull data network architectures employing multiple bridged peripheral links can be used to create an Ultra-Fast, Ultra-Secure, Private and Low Power data network capable of connecting nearly any system across multiple connection mediums. Bridging standard USB 3.0 technologies, we demonstrate a universally secure and scalable switchable data platform offering the highest level of data privacy, security and performance. Further supporting heterogeneous connections across different protocols, data may flow easily across the entirety of available consumer electronics systems with unparalleled security, privacy and transfer speed.
- 13:50 A Hardware-Trojan Classification Method Utilizing Boundary Net Structures
- Recently, cybersecurity has become serious concern for us. In this paper, we propose a hardware-Trojan classification method utilizing boundary net structures. To begin with, we use a machine-learning-based hardware-Trojan detection method and classify the nets in a given netlist into a set of normal nets and that of Trojan nets. Based on the classification, we investigate the nets around the boundary between normal nets and Trojan nets and extract the features of the nets identified to be normal nets or Trojan nets mistakenly. Finally, using the classification results of machine-learning-based hardware-Trojan detection and the extracted features of the boundary nets, we classify the nets in a given netlist into a set of normal nets and that of Trojan nets again. The experimental results demonstrate that our method outperforms an existing machine-learning-based hardware-Trojan detection method in terms of true positive rate.
- 14:10 A CAPTCHA-like Implicit Authentication Model for Smart TV Based on Machine Learning Algorithms
- Recently, IoT (Internet of Things) environment centered on Smart TV has been developed exponentially in order to provide a convenient and connected life experience to users. But this environment is vulnerable to a hacking attempt, as each of the devices is exposed to the hacker. To deal with trade-off between security and convenience, this paper proposes an implicit authentication method based on users' TV viewing history. A machine learning model is trained on the server with face set, and it is operated on the client TV after deployment. When compared to previous authentication methods, our model has an advantage in user convenience, and it is robust in terms of security as it updates the model regularly.
- 13:30 Utilization of Two Microphones for Real-Time Low-Latency Audio Smartphone Apps
- This paper presents an approach to overcome the limitation imposed by existing smartphone operating systems in using two microphones of smartphones at the same time for real-time low-latency audio apps that require the use of two microphones. This approach involves the use of an external dual-microphone and the processing steps needed to access audio signals from both of the external microphones at the same time. The developed approach is then applied to two example audio applications consisting of noise classification and noise reduction. It is shown that these dual-microphone apps lead to improved noise classification and noise reduction accuracy compared to the apps using a single microphone.
- 13:50 Image Placement Order Optimization for Mobile Commerce
- With the increase in popularity of mobile commerce, the placement order of images on mobile displays is an important factor to attract and retain customers' attention. In this paper, we propose a novel method to optimize the placement of item images based upon the relative attractiveness of each image to the customer. To judge this, the proposed method estimates the leave rate of the customer at each image for each ordering using a model based on unsupervised hierarchical clustering. This allows estimating the expected leave rate for different image placements, solving an optimization problem to obtain the best ordering. The model is evaluated using a dataset collected from Rakuten Ichiba, the largest e-commerce site in Japan.
- 14:10 Saving Power Consumption of Smartphones in the Screen-off State with Disabling the Wi-Fi
- Smartphone OSs have a function with which applications can be invoked without users' operation. These invocations heavily consume battery. Temporal disabling network device is a promising methods for decreasing power consumption in the screen-off state. Disabling decreases power consumption, but processes of disabling and enabling the device consume power. In our work, we focus on a method for reducing power consumption in the state by repeating to disable and enable the network device and propose methods for estimating Break-Even Time (BET). We them evaluate the methods and show that the method can correctly estimate BET.
ST05-2: Architecture-Level Energy, Security, and Reliability Solutions for CE Digital Hardware (ESR) - Session 2
- 13:30 Facial Biohashing Based User-Device Physical Unclonable Function for Bring Your Own Device Security
- Bring your own device (BYOD) is gaining popularity. Using multifarious personal devices in the workplace to perform work-related tasks brings new challenges to trust and privacy management. Existing authentication schemes usually target at user or device separately, while the BYOD system needs to ensure that only the authorized user with the trusted devices can be given access. This paper presents a novel biohashing based user-device physical unclonable function (UD PUF) to provide a bipartite authentication of both user and device for the BYOD system. Biometric features are extracted as user identity while PUF endows the device with an inseparable and unclonable ``fingerprint". Biohashing acts as an intermediary between these two incoherent macroscopic biometric and microscopic silicon entropy sources for security enhancement. The concept is demonstrated using a 64x64 image sensor PUF simulated in 180 nm 3.3 V CMOS technology, and the ORL and yale databases of faces. Our preliminary experimental results showed that a genuine (user, device, challenge) combination exhibits a very low equal error rate of 0.032, and tampering of any elements of the tuple will cause the hamming distance between the "live" and enrolled templates to have nearly random distribution.
- 13:45 Study of Hardware Trojans Based Security Vulnerabilities in Cyber Physical Systems
- The dependability of Cyber Physical Systems (CPS) solely lies in the secure and reliable functionality of their backbone, the computing platform. Security of this platform is not only threatened by the vulnerabilities in the software peripherals, but also by the vulnerabilities in the hardware internals. Such threats can arise from malicious modifications to the integrated circuits (IC) based computing hardware, which can disable the system, leak information or produce malfunctions. Such modifications to computing hardware are made possible by the globalization of the IC industry, where a computing chip can be manufactured anywhere in the world. In the complex computing environment of CPS such modifications can be stealthier and undetectable. Under such circumstances, design of these malicious modifications, and eventually their detection, will be tied to the functionality and operation of the CPS. So it is imperative to address such threats by incorporating security awareness in the computing hardware design in a comprehensive manner taking the entire system into consideration. In this paper, we present a study in the influence of hardware Trojans on closed-loop systems, which form the basis of CPS, and establish threat models. Using these models, we perform a case study on a critical CPS application, gas pipeline based SCADA system. Through this process, we establish a completely virtual simulation platform along with a hardware-in-the-loop based simulation platform for implementation and testing.
- 14:00 Re-envisioning Digital Architectures Connecting CE Hardware for Security, Reliability and Low Energy
- Push-Pull data network architectures employing multiple bridged peripheral links can be used to create an Ultra-Fast, Ultra-Secure, Private and Low Power data network capable of connecting nearly any system. Initially implemented using standard USB technologies, crossPORT demonstrates a universally secure and scalable switched data platform offering the highest level of data privacy, security and performance. A "Software Constructed" data network, the presented architecture offers frictionless adoption and price-performance measures that exceed 10X. Further supporting heterogeneous connections across different protocols, data may flow easily across the entirety of available consumer electronics systems with unparalleled security, privacy and transfer speed.
- 14:15 Energy-Recovery Based Hardware Security Primitives for Low-Power Embedded Devices
- Hardware based security vulnerabilities such as side-channel attacks, IC piracy, IC counterfeiting, etc. are major threats for the successful deployment of the embedded computing systems. Further, the embedded computing devices work on a limited power supply (battery operated devices), and therefore have stringent constraints on power consumption. In this paper, we propose Energy Recovery (ER) computing as a novel platform to design low-power hardware security primitives for embedded computing devices. As a case study, we have discussed the design of two important ER-based hardware security primitives: (i) low-power and Differential Power Analysis (DPA) resistant PRESENT cryptographic core, (ii) low-power Physically Unclonable Function (PUF). Through simulations, it is illustrated that the ER computing based hardware security primitives are low-power as compared to the existing approaches. For example, the PRESENT-80 cryptographic algorithm implemented using ER computing consumes 45% less power as compared to its CMOS implementation. Further, ER computing in emerging devices such as FinFET and Tunnel FET to design low-power hardware security primitives is also illustrated.
- 13:30 An Energy Efficient Epileptic Seizure Detector
- Epilepsy is one of the most common neurological disorders affecting up to 1\% of the world's population and approximately 2.5 million people in the United States. Seizures in more than 30\% of epilepsy patients are resistant to anti-epileptic drugs. A significant focus of current research efforts is the development of an energy efficient implantable integrated circuit for real-time detection of seizures. In this paper we propose an architecture for an implantable seizure detector using a hypersynchronous signal detection circuit and signal rejection algorithm (SRA). The proposed seizure detector (SD) continuously monitors neural signals for hypersynchronous pulses and extracts the seizure onset signal. If the pulses in an epoch exceed a threshold value, a seizure is declared. The design was validated using \simulink. The signal rejection algorithm (SRA) reduces false detection and minimal circuitry leads to a 12\% reduction of power consumption.
- 13:50 A Human Activity Recognition-Aware Framework Using Multi-modal Sensor Data Fusion
- Recent years, with smartphones and other pervasive devices the paradigm of situation-recognition extended to IoT devices in home. Since IoT devices produce data that can help predict accidents or disasters in private or public environments. In IoT devices provided situation, machine learning technologies can help make an insight to what's really meaningful according to service provider's purpose. So, in order to understand several kinds of user's situation within an environment with IoT devices, a novel human activity recognition scheme is required to manage lots of data for guaranteeing accurate situation in real-time. As the accuracy of analyzing result and the response time of informing a situational corresponding to users are important factors on providing services, we present a human activity recognition framework based on home gateway connected with IoT devices and consumer devices in home for consumer services.
- 14:10 Project OurPuppet: A System to Support People with Dementia and Their Caregiving Relatives at Home*
- This paper discusses the current state of project "OurPuppet" which is a system to support elderly people suffering from dementia. In contrast to other system, OurPuppet not only focus on the patient but also on relatives who are taking care of the elderly. Taking care for people with dementia is a high burden for caregiving relatives. If the caregiving relatives aren't present, many dependent persons show a high insecurity. In turn, this limits the freedom of caregiving persons and hence leads to an emotional burden or even to an isolation of the caregiving relatives. In the Project OurPuppet we are developing a sensor based interactive puppet. The puppet is supposed to support the caregiving relatives, relieve the burden from them and create new free spaces for their own social life. It acts as a companion for the dependent people as well as a communication interface between the caregiving relatives and them. To this end, the puppet is equipped with suitable sensors to detect and monitor the emotional condition of the depending person and provide the caregiving relatives with the necessary information. The puppet can also show emotional expressions in order to provide human like interaction capabilities. The puppet can intervene in situation where the dependent person needs assistance e.g. it can remind them that their caregiving relatives come back soon. This shall prevent isolation and decrease emotional burden of the caregiving relatives. Finally it will improve the situation for the dependent people as well.
- 13:30 Secure Customer Data over Cloud Forensic Reconstruction
- Cloud computing has been evolving over the last couple of years. Adoption of the cloud services for usage in real time has been increasing ever since in the era of Big Data and Internet of Things (IoT). Adoption of Cloud is evident and can not be stopped by the individuals and small organizations. It offers many features, at low or reduced cost and provides the great potential. Cloud computing provides that extra support to move applications from the traditional local systems to cloud service provider's storage locations. But there are many security aspects which need to be considered while moving to Cloud, which are yet be solved to be completely secure in the Cloud. In this paper, we look into new security issue in the Cloud forensics point of view. Can we secure the data over cloud forensic reconstruction of the private data once that customer has done with the services and taken out their private data? The main focus of this paper is to look for an alternative ways to protect the data in Cloud.
- 13:42 Privacy-Preserving Context-Aware Friend Discovery Based on Mobile Sensing
- With the advancement of smart phones, wearable devices and communication technologies, it becomes very convenient to obtain various types of data through mobile sensing, where data can be sensed via embedded sensors, stored, processed, and transmitted to anywhere. Mobile sensing has been studied extensively in both academy and industry, and has been applied in many novel applications in the past few years, such as environment and traffic monitoring and mobile health. In this work, we propose a new application scenario called context-aware friend discovery based on mobile sensing, where contextual attributes such as location, weather and temperature are utilized to enhance the existing friend discovery schemes. However, data privacy becomes a major concern for consumers to accept this application. To address this issue, we further propose a privacy-preserving context-aware friend discovery scheme, where a user's sensitive data is well protected. Security analysis shows the correctness and privacy guarantee of the proposed scheme.
- 13:54 A Hybrid Cloud and Fog Architecture for Vehicle Behavior Understanding
- As the unprecedentedly rapid development of urbanization nowadays, it becomes more and more significant to build a comprehensive understanding of our urban elements. However, being an important part of our cities, vehicle behavior learning is still a big challenge in Smart Cities since the existing computing platforms are not able to achieve the goal that vehicle behavior can be learned in real-time with high accuracy. In this paper, we propose a hybrid computing platform for urban vehicle behavior analysis in which Cloud Computing is utilized for knowledge base construction with long-term and large scale trajectory datasets and Fog Computing is used for behavior analysis at the edge of network. Furthermore, case based reasoning (CBR) is adopted for understanding the behaviors given new vehicle trajectories at the Fog layer. The experimental results prove the feasibility of our proposed computing platform and vehicle behaviors can be analyzed in real-time with high accuracy.
- 14:06 A Scalable Purchase Intention Prediction System Using Extreme Gradient Boosting Machines with Browsing Content Entropy
- Nowadays, a prosperity of electronic commerce (E-commerce) not only provides more convenience to consumers but brings more new opportunities in online advertising and marketing. Online advertisers can understand more about consumer preferences based on their daily online shopping and browsing behaviors. The development of big data and cloud computing techniques further empower advertisers and marketers to have a data-driven consumer-specific preference recommendation based on the online browsing histories. In this research, a decision support system is proposed to predict a consumer purchase intention during browsing sessions. The proposed decision support system categorizes online browsing activities into purchase-oriented and general sessions using extreme boosting machines. With the browsing content entropy features, the proposed method achieves 41.81% recall and 34.35% F1 score. It further shows its strong predictive capability compared to other benchmark algorithms including logistic regression and traditional ensemble models. The proposed method can be implemented in real-time bidding algorithms for online advertising strategies. Ad deliveries on browsing session with potential purchase intention not only improve the effectiveness of advertisements, but significantly increase last touch attributions for campaign performance.
- 14:18 Wait Analysis of Virtual Machines Using Host Kernel Tracing
- An agent-less method to understand virtual ma- chines (VMs) behavior its evolution during the VM life-cycle is an essential task for IaaS provider. It allows the IaaS provider to better scale the VMs resources by properly allocating the physical resources. On the other hand, because of privacy, security, ease of deployment and execution overhead issues, the method presented limits its data collection to the physical host level, without internal access to the VMs. We propose a host-based, precise method to recover wait states for the virtual CPUs (vCPUs) of a given VM. The Wait Analysis Algorithm (W2A) computes the state of vCPUs through the host kernel trace. The state of vCPUs is displayed in an interactive trace viewer (TraceCompass) for further inspection. Our proposed VM trace analysis algorithm has been open-sourced for further enhancements and to the benefit of other developers. Our new technique is being evaluated with representative workloads, generated by different benchmarking tools. These approaches are based on host hypervisor tracing, which brings a lower overhead (around 0.03%) as compared to other approaches.
Friday, January 12 14:30 - 16:00
- 14:30 Field Experiment of Hybrid Video Delivery Using Next-Generation Terrestrial Broadcasting and a Cellular Network
- The utilization of hybrid video delivery is important requirements for next-generation terrestrial broadcasting. This is especially true for mobile reception because the mobility of the user changes the receiving condition. We implement a hybrid video delivery system using terrestrial broadcasting and a cellular network to maintain user experience.
- 14:52 Spectral Efficiency Gains Through the Use of FBMC/OQAM for DOCSIS Systems
- DOCSIS provides internet access via HFC networks. With the latest version DOCSIS 3.1, the Physical Layer changed from SC-QAM to OFDM. Recent activities in the 5G development investigated other modulation schemes beyond OFDM. One candidate was FBMC, often used in combination with OQAM. FBMC promises fewer out-of-band emissions and no overhead due to the Cyclic Prefix. This paper introduces simulations that compare the FBMC/OQAM system to the DOCSIS 3.1 downstream in terms of robustness and spectral efficiency. Simulations show that the FBMC/OQAM system can improve the spectral efficiency by 10 % while providing similar robustness for HFC network channels.
- 15:15 Constructing Universal Secure Multi-Medium Data Networks Across Consumer Systems
- A novel Push-Pull data network architecture is hereto presented, employing multiple bridged peripheral links to create an ultra-fast, ultra-secure, private and low power data network to connect nearly any system. Bridging standard USB 3.0 technologies, we demonstrate a universally secure, ultra-low power and scalable switchable data platform offering the highest level of data privacy, security and performance. Delivering up to 12 times the throughput speeds of existing USB 3.0 data transfer cables, the presented solution builds on the reliability of universal peripheral communications links using proven ports, protocols and low-power components. A "Software Constructed" ad-hoc circuit network, the presented digital architecture delivers frictionless adoption and exceptional price-performance measures connecting both existing and future systems.
- 15:37 Performance Evaluation of Dual-Polarized MIMO Ultra-Multilevel OFDM Using NU-QAM Under Mobile Reception
- The combination of dual-polarized MIMO and ultra-multilevel OFDM is considered in the next-generation digital terrestrial broadcasting. Also, the use of Non-Uniform QAM(NU-QAM) as a new sub-carrier modulation scheme has been proposed. In the mobile reception of OFDM signal, inter-carrier interference(ICI) is generated by Doppler-spread. MIMO-ICI canceller using iterative detection can improve the reception characteristics with complexity reduction. In this paper, the reception characteristics are evaluated for dual-polarized MIMO ultra-multilevel OFDM using NU-QAM under mobile reception.
- 14:30 Monitoring System for a Single Aged Person on the Basis of Electricity Use- Heatstroke-Prevention System -
- In order to develop monitoring systems for single aged persons by grasping their usage of electricity, especially by using smart meters prevailing recently, we are hatching out several algorithms to infer whether residents are living ordinary lives or not. This paper mentions a method to prevent heatstroke inside a home as one of the algorithms. The system estimates an on-off condition of an air-conditioner in the home by using data provided by the smart meter. If the outdoor air temperature is severely high and the air-conditioner is inferred not being turned on, the system will raise an alarm.
- 14:48 Design of the Intuitive Touch Screen and the Camera Control of a Remote Video Conference System
- To provide a more convenient operation, this design uses a popular Tablet PC touch screen function, an intuitive touch positioning algorithm (ITPA), and a video conference system based on the Session Initiation Protocol (SIP). By utilizing this design the M.D. can easily access the remote camera control of the remote video conference system in order to examine the different parts of the patient's body.
- 15:06 A Real-Time Sleeping Position Recognition System Using IMU Sensor Motion Data
- The emergence of wearable miniature inertial measurement unit (IMU) sensors is a powerful enabler for lying motion data extraction. In this study, an IMU sensor is used for capturing 3D motion data. A spectrogram based algorithm for feature extraction from the motion data is proposed and implemented. Using the generated spectrogram based features, the long term short memory (LSTM) recurrent neural network (RNN) model is used for recognition of sleeping positions. The test results show that an accuracy of 99.09% can be achieved in a supervised learning mode. A real-time feature extraction and recognition system is developed to implement the proposed algorithm.
- 15:24 Improving Mobility and Autonomy of Disabled Users via Cooperation of Assistive Robots
- Robotic assistance for disabled users and/or for elderly people represents at the same time an active research area and a growing market, which will lead to commercial technological solutions in a near future. Cooperation of existing assistive technologies thus represents an important step towards the development of such solutions. In this paper, we face the problem of cooperation between a smart wheelchair and one or more robotic workstations: the wheelchair can navigate autonomously and permits to choose the point to move to or the object to interact with, while the robotic workstation can interact with one or more objects within its workspace and with the user on the wheelchair. The hardware/software architecture is here presented, together with some preliminary results.
- 15:42 Force and Activity Monitoring System for Scoliosis Patients Wearing Back Braces
- Monitoring the effectiveness of brace treatment of scoliosis is an ongoing challenge that many physicians face today. Scoliosis is a medical condition which occurs in adolescents, where an individual's spine develops curvature. To successfully monitor compliance with brace treatment, a wearable multi-modal sensor solution is embedded into the patient's brace. The custom-designed hardware consists of a sensor board, a force sensor, an accelerometer and a gyroscope. The force sensor collects the force being exerted on the patient's back, while the accelerometer and gyroscope generate cues to determine the patient's activities and lifestyle. In this paper, we propose a novel embedded ubiquitous system to identify patient activities and evaluate the effectiveness of the brace treatment pervasively. Data is collected through the sensors and stored in a SD memory for longitudinal analysis. Our experimental results on real patients demonstrated that we achieved an overall accuracy of a 100% for activity detection in a semi-supervised experimental setup. It was also observed that the duration of brace wear increased from 20% to 80% in 4 weeks.
- 14:30 ANN-based Stride Detection Using Smartphones for Pedestrian Dead Reckoning
- Position awareness is a very important issue for internet of thing (IoT) applications using smartphones. Pedestrian dead reckoning (PDR) is one of the methods used to estimate a user's indoor position. The accuracy of a stride detection is very important to guarantee the estimation accuracy of the user location. This paper proposes an algorithm to detect the stride using acceleration spectrogram feature by utilizing the accelerometer in a smartphone. An artificial neural network (ANN) technology is applied to detect the stride. The proposed algorithm has an accuracy of 97.7% for stride detection.
- 15:00 Lazy TRIM: Optimizing the Journaling Overhead Caused by TRIM Commands on Ext4 File System
- We propose a novel scheme, called lazy TRIM to reduce journaling overhead caused by the TRIM command.
- 15:30 Smartphone Based Lifelog with Meaningful Place Detection
- With the advancement of electronic techniques, current smartphones can record location information with the help of WIFI signal, base station signal and Global Positioning System (GPS) receivers embedded in them. A smartphone can record where we go and where we stay. However, all the locations are recorded using latitude and longitude, which make it hard for users to have an intuitive feelings of these data. In this paper, we propose a way to find places where users stay for a while to perform some activities with the location data tracked by smartphone, and use a cascade classifier to infer users' different activities in different places. With places and related activities, we can build a lifelog for smartphone users.
- 14:30 Optimized Vision-Directed Deployment of UAVs for Rapid Traffic Monitoring
- Traffic monitoring through conventional sensor can be expensive, whereas using UAVs can be both cost-efficient and more flexible. In this paper, we a framework for rapid deployment of UAV-based systems for traffic monitoring. The framework considers a computer vision pipeline for vehicle density estimation, as well as the characteristics of the urban area with the goal of determining a-priori the optimal locations for a UAV in order to monitor an area, thus facilitating fast deployment. Using the city of Nicosia as a benchmark we show that our approach can find the optimal location for the placement of UAVs.
- 14:52 Drones as Collaborative Sensors for Image Recognition
- Distributed sensor networks have the ability for more data collection and advanced tasks such as object recognition and tracking than a singular sensor. However, these sensor networks are often limited in their ability to coordinate either by their requirement to be connected to a central system or by their fixed position and static nature. Drones or unmanned aerial vehicles (UAVs) provide a unique opportunity in multi-sensor networks. Although drones are usually limited by their available power due to battery capacity, recent advances in technology and algorithms provide drones with more available computational power. In addition to computability augmentation in drone mobility, decentralized coordination enables drones to create the next generation of distributed multi-sensor networks. In this paper, we explore combination of the existing image processing and object recognition techniques in the perspective of collaborative drones, which can improve the robustness of image recognition tasks.
- 15:15 Physiological Characterization of Need for Assistance in Rescue Missions with Drones
- The use of drones is recently gaining particular interest in the field of search and rescue. However, particular skills are still required to actively operate in a mission without crashing the drone. This limits heir effective and efficient employment in real missions. Thus, to assist the rescuers operating in stressful conditions, there is a need to detect an increase of workload that could compromise the outcome of the mission. In this work a simulator is designed and used to induce different levels of cognitive workload related to search and rescue missions. Physiological signals are recorded and features are extracted from them to estimate cognitive workloads. The NASA Task Load Index is used as subjective self-report workload reference. Then, performance is recorded to objectively evaluate the execution of the tasks. Finally, the analysis of variance (ANOVA) is used to verify intra- and inter-subject variability. Results show statistical decrease of the mean normal-to-normal (NN) interval with an increase of cognitive workload. Moreover, it is observed a decrease of performance while an increase of cognitive workload exists. This information can be used to detect the need for assistance.
- 15:37 Base Station Positioning for Industrial Wireless Sensor
- Wireless Sensor Networks (WSN) have increasingly been employed in a wide range of industrial applications mainly for monitoring, quality control and diagnosis of the industrial equipment and machines. In such industrial applications, one of the most challenging problem is to determine an optimal positions for mounting the wireless sensor nodes for the highest quality in signal reception. It is typically assumed that the sensor nodes used for monitoring the industrial machines are deployed on fixed locations adjacent to the machines, finding the position of the base station is an important problem that requires attention. In most cases, due to the harsh conditions of the industrial environment, it is necessary to utilize more than one base stations to improve the signal quality, which makes the problem more complicated. In this paper, we focus on the problem of estimating the number and optimal position for wireless base stations in an industrial application of WSNs. We present a k-means clustering based method for estimating not only the optimal position, but also the optimal number of the base stations. The simulation result illustrate that increasing the number of the base station to some extend will enhance the communication link quality in industrial environments where there is a huge effect of fading and shadowing
Opening Address: H. Okumura (Toshiba, Japan)
- 14:30 Automotive Displays from Direct View to AR Head-Up
- Advanced displays enable new multimodal interactions for automotive use like (semi-) autonomous driving ranging from direct view to AR head-up displays. This paper presents selected challenges and solution for future automotive applications.
- 14:52 Human Interface Based on an Interaction Between Driver's Capability and Task Demands
- This paper describes human interface design methods for in-vehicle systems. Task-Capability Interface model suggests a concept of safe driving based on an interaction between driver's capability and task demands. Drivers proceed the driving task safely when their capability (that was influenced mainly by their physical and cognitive conditions) exceeds the task demands (that are changed by the road traffic conditions and their driving operations). The paper focuses on an estimation method of the task demands and interface design factors for the estimated visual demand. Finally, we propose human interface of the automated driving systems based on the Task-Capability Interface model.
- 15:15 Proposal of Next-Generation HMI Utilizing Context and Personality Theories
- The relationship between humans and cars is changing; cars are increasingly working autonomously instead of being operated by humans. However, humans tend to experience anxiety, annoyance, overconfidence or distrust when machines act against human expectations. Therefore, Human-Machine Interface (HMI) needs to evolve from a design aiming at automotive control to a logical design that intends to bring about cognitive-behavioral changes to meet human expectations. In this presentation, we propose 'Next-Generation HMI' that utilizes context and personality theories for understanding human expectations and prompting changes in human behavior.
- 15:37 Verification of HMI for an Automated Driving System by Using a Driving Simulator
- Automated vehicles with automation level of 3 will be launched by 2020 from automotive companies. This paper introduces our activity to design and verify human-machine interface (HMI) to safely realize the driving delegation between the driver and the ADS. The system model of ADS designed is utilized to build a driving simulator to be used for HMI verification.
Friday, January 12 16:00 - 16:30
- 16:00 Biometrics-as-a-Service: A Framework to Promote Innovative Biometric Recognition in the Cloud
- Biometric recognition, or simply biometrics, is the use of biological attributes such as face, fingerprints or iris in order to recognize an individual. A key application of biometrics is authentication; i.e., using said biological attributes to provide access by verifying identity. This authentication provides security and privacy for the hardware or the software systems. This paper presents a framework for Biometrics-as-a-Service (BaaS) that performs recognition operations in the cloud, while relying on simple and ubiquitous end-user devices such as smart-phones. Further, the framework promotes innovation by providing interfaces for a plurality of software developers to upload their recognition algorithms to the cloud. When a biometric authentication request is submitted, the system uses a criteria to automatically select an appropriate matching algorithm. Every time a particular algorithm is selected, the corresponding developer is rendered a micropayment. This creates an innovative and competitive ecosystem that benefits both software developers and end-users. As a case study, we have implemented the following: (a) an ocular recognition system using a mobile web interface providing user access to a biometric authentication service, and (b) a Linux-based virtual machine environment used by software developers for algorithm development and submission.
- 16:01 Downloadable Trusted Applications on Tizen™ TV TrustWare™ Extension: As a Downloadable Application Framework
- This article describes Tizen TV's new features on downloadable trusted application supports. It allows developers to write, build, debug their trusted applications on top of Tizen TV's REE(rich execution environment) and TEE(trusted execution environment). To make premium contents distribution more secure, Tizen TV also provides DRM Intrinsic TA(trusted application) to support joining secure media pipeline. Without security risks, service providers can take advantage of this downloadable TA development as well as the DRM Intrinsic TA on interoperability with their existing protection systems, and Tizen TV products don't need to upgrade the firmware for adding trusted applications any more.
- 16:03 PRNU-based Source Identification for Network Video Surveillance System
- Photo response non-uniformity (PRNU) noise is instrumental for camera source identification in image forensics. The paper proposes a signal-based detection system using PRNU for source identification in video surveillance systems. The effects of different aspects such as video resolutions, frame types and environmental conditions on the accuracy and reliability of the system have been tested. Our results show that the signal-based approach is effective to verify the video source.
- 16:04 Design of an LPWAN Communication Module Based on Secure Element for Smart Parking Application
- LPWAN is expected to be used in various IoT applications such as smart parking, environment monitoring, etc. This paper introduces a method of processing LoRaWAN protocol which is one of LPWAN using Secure Element. If important information such as keys are stored in a General purpose MCU which is used frequently by IoT device, keys can be easily leaked by hacking. Therefore, this paper proposes a method to safely store keys and to process LoRaWAN protocol more securely based on Secure Element. In this paper, security functions such as data encryption and message signing are handled by the Secure Element. Therefore, the keys required to perform these functions are also managed within the Secure Element. And, this paper also proposes efficient integrity verification method of the executable of the main MCU based on Secure Element.
- 16:06 White List-based Ransomware Real-time Detection and Prevention for User Device Protection
- A ransomware encrypts valuable user/system files in victims, and then asks for a ransom to release a decryption key. So far, ransomware countermeasures are very similar to existing malware countermeasures which use patch services to distribute malware information. However, such approaches cannot prevent attacks of new/variant ransomwares. To detect/control ransomwares, this paper proposes to apply an access control scheme to the file operation procedure of the operating system on user's device. Since the proposed scheme does not use the patch of a ransomware information published by a trust party, it can prevent both new and variant ransomwares in real time.
- 16:07 Effect of Hardware Trojans on the Performance of a Coded Communication System
- Pernicious Trojan circuits inserted during the design or fabrication phases of an integrated circuit may cause undesirable effects in the designed system. In the case of communication systems, incorrect decoding of the received information is an issue that may lead to serious consequences. In this paper, we study the performance degradation of a coded communication system, due to the presence of hardware Trojans in the decoding unit. It can be seen that the Trojans have negligible effect at low SNRs while there is a statistically significant increase in the bit error rate (p < 0.01) at higher SNRs. At high SNRs, we expect the BER to tend towards zero, but it is observed that even with the Trojan being activated only once during the entire duration of the transmission, there is a constant bit error rate of the order of 0.0001.
- 16:09 Using NFC Tags and Smartphones to Design a Reliable Mechanism to Pick a Child Up from School
- When school is over, parents picking up their child can be disoriented. This design provides a convenient mechanism to identify either the parents or any authorized person who can pick up their child. This design uses the smartphone NFC function and NFC tags to identify the authorized individual. This design provides a simple user interface to manage the school class table of children who are to be picked up. When either the parents or the authorized persons want to pick up their child, the teacher's smartphone scans the parents' or authorized individual's NFC tag and verifies the NFC tag with this table of children who are to be picked up within a short time.
- 16:10 Efficient and Secure Device Clustering for Networked Home Domains
- Virtual private community (VPC) architecture efficiently and securely creates personal domains to share content only in the networked domains without external authentication/proxy systems. For that, a secure device clustering method consisting of a device identification and membership management system is required. However, existing certificate chain based association (CCA) scheme for VPC, one of device clustering schemes, has two unsolved problems. First, since it does not verify who is an original creator of a VPC domain, it makes VPC vulnerable to spoofing attacks. Second, it does not consider how to restrain revoked members from accessing VPC domains, which causes malicious users to illegally access VPC domains. Hence, to solve such problems, this paper proposes a member list chain and reputation based association (MLC-RbA) scheme, and then evaluates the performance/security of MLC-RbA to prove that MLC-RbA provides better reliability
- 16:12 Malware Propagation Effects on SCADA System and Smart Power Grid
- Critical infrastructures have suffered from different kind of cyber attacks over the years. Many of these attacks are performed using malwares by exploiting the vulnerabilities of these resources. Smart power grid is one of the major victim which suffered from these attacks and its SCADA system are frequently targeted. In this paper we describe our proposed framework to analyze smart power grid, while its SCADA system is under attack by malware. Malware propagation and its effects on SCADA system is the focal point of our analysis. OMNeT++ simulator and openDSS is used for developing and analyzing the simulated smart power grid environment.
- 16:13 DDefender: Android Application Threat Detection Using Static and Dynamic Analysis
- Android is the most widespread mobile operating system in the world. Due to its popularity, malware has been increasing every year steadily, which causes lots of problems to users, such as using the device's resources and transmitting private information without user's awareness. As malware has increased, anti-malware solutions have as well. Current anti-malware solutions often have very serious limitations and malware is becoming more apt to take advantage of them. In this paper, we present DDefender, a user-friendly application that detects Android malicious applications on device. DDefender is a comprehensive solution that utilizes static and dynamic analysis techniques to extract features from the user's device, then applies deep learning algorithm to detect malicious applications. At first, we use dynamic analysis to extract system calls, system information, network traffics, and requested permissions of an inspected application. Then we use static analysis to extract significant features from the inspected application such as application's components. By utilizing neural network and a large feature set of 1007 features, we evaluated our system with 4208 applications (2104 benign applications and 2104 malicious applications) and we achieved up to 95\% accuracy.
- 16:15 Real-time Eye Tracking for Password Authentication
- Personal identification numbers (PINs) are widely used for user authentication and security. Password authentication using PINs requires users to physically input the PIN, which could be vulnerable to password cracking via shoulder surfing or thermal tracking. PIN authentication with hands-off gaze-based PIN entry techniques, on the other hand, leaves no physical footprints behind and therefore offer a more secure password entry option. Gaze-based authentication refers to finding the eye location across sequential image frames, and tracking eye center over time. This paper presents an application for real-time gaze-based PIN entry and eye tracking for PIN identification using a smart camera. Captured data is processed to determine the digits gazed by the user and their order.
- 16:16 Energy-Aware Core Switching for big.LITTLE Multicore Mobile Platform
- This paper addresses the problem of the core selection in heterogeneous multicore processors with big.LITTLE architecture used for mobile platforms. It is shown that core switching for less energy consumption must consider both core frequency and utilization.
- 16:18 Load Monitoring Effects and Characteristics of DC Home
- Recently most of the appliances are using DC power system internally, and DC based home has been proposed pursuing energy efficiency. The load monitoring is important for home electric energy management to identify energy consuming status. However, conventional approaches have been based on AC power systems. In this work, the effects of load monitoring at DC home are studies. Based on the data measured through a DC home power distribution testbed, the characteristics of DC home are discussed. The results show the effectiveness of load monitoring in DC home environments.
- 16:19 An Automatic Temperature Control for Induction Cooktops to Reduce Energy Consumption
- Induction cooktops are very used as an efficient alternative to traditional cooking systems such as gas hobs. Even if the energy efficiency of induction cooktops is twice as much as the traditional gas burners, the differences in terms of the energy price can limit the diffusion and marketing of induction cooktops. Recent eco-normative are regulating the energy consumption of this kind of household appliances sold in different countries. This paper proposes an automatic temperature control to reduce the energy consumption for induction cooktops. As test case, the water boiling test was used to simulate and validate the proposed approach.
- 16:21 Infrared-Camera-based Metamer Marker for Use in Dark Environments
- We propose an invisible metamer marker that can be used in low illumination environments. The proposed marker is not visible to the eye as the color of the background is similar. In our system, the marker and the background image look differently appeared in the infrared camera image by using different printers for each region. The proposed method is able to produce uniform quality unlike existing invisible markers. Based on these metamerism features in infrared band, markers can be successfully used even in low light and dark environments and the commercial applicability is high.
- 16:22 A Practical Digital Door Lock for Smart Home
- Digital door lock plays an important role on the smart home system, not only for door guardian but also for family member incoming/outgoing awareness. However, most of the current digital door locks still keep the mechanical keyhole with digital interface to fulfill the traditional habit of key usage and consequently results in the doubt of redundant interface design. For this phenomenon, we discuss the design of digital lock and propose a mechanism to keep the traditional key usage with the digital interface. Moreover, We develop a smart-lock centered service model and create the possibility of service orchestration with the smart home system. We believe that the proposed digital door lock can survival in the power failure and provide more creative living experience.
- 16:24 Research on Testing Method of Heterogeneous Software Based on CDD Model
- The complicated heterogeneous software that is featured by systems such as: E-commerce, E-government affairs, enterprise informatization, has developed at a rapid rate. In order to guarantee the quality of heterogeneous software, it is an urgent issue to research the testing method that is suited for its features. The software test based on model can generate test case according to software behavior model and structure model, effectively improve test efficiency. This article, based on the analysis of the feature of heterogeneous software, proposes test model of heterogeneous software—CCD (Component Collaboration Diagram) model, by constructing Components Specification Framework, and CCD to model the interactive relations between heterogeneous components. We put forward the testing method of heterogeneous software system.
- 16:25 Sparse Audio Inpainting with Variational Bayesian Inference
- Audio inpainting is defined as the process of restoring the damaged segments of an audio signal, based on the known signal values and prior information about the signal. In this paper, we formulate the problem in a Bayesian framework and adopt an efficient sparsity inducing Student's-t prior distribution, assumed for the discrete cosine transform coefficients, applied on the signal. We also propose a variational Bayesian algorithm for inpainting, that performs approximate, though tractable, inference. Lastly, experiments demonstrate the efficiency of the proposed methodology when used for declipping audio signals, by comparing with the state-of-the-art.
- 16:27 Gaussian Process Regression for Single-channel Speech Separation
- Gaussian process (GP) is a flexible kernel-based learning method which has found widespread application in signal processing. In this paper, a supervised method is proposed for handling single-channel speech separation (SCSS) problem. In this work, we focus on modeling a nonlinear mapping between mixed and clean speeches based on GP regression, in which reconstructed audio signal is estimated by the predictive mean of GP model. The nonlinear conjugate gradient method was utilized to perform the hyper-parameter optimization. An experiment on a subset of TIMIT speech dataset is carried out to confirm the validity of the proposed approach.
- 16:28 A Novel Secure Simple Bluetooth Pairing Using Physical Vibration
- We propose a novel secure simple Bluetooth pairing technology using physical vibration communication. We describe the modulation and demodulation of the vibration communication and propose a new Bluetooth secure simple pairing. Also, we confirmed the data rate of the vibration communication to be up to about 40 bps on a smartphone.
Friday, January 12 16:30 - 17:30
Panelist: (1) Robert Bohn, NIST, Cloud Computing Program Manager, Advanced Networking Technologies Division, Information Technology Laboratory (ITL) (2) Donald R Deutsch, Oracle, Vice President, Chief Standards Officer (3) Joel Fleck, CA Technologies, Principal, Industry Standards and Open Source, Office of the CTO (4) John Messina, National Institute of Standards and Technology (NIST), Supervisory Computer Scientist
- 16:30 Cloud Federation Standards
- Panel Title: Cloud Federation Standards Panel Keywords: Cloud; Federation; Standards; Federated Cloud Track: ST02 Expert Panel Proposals (EPP) Panel Theme: There is growing recognition that the lack of cloud federation in a landscape of multiple independent cloud providers is limiting the service reach, resources and scalability that can be offered in a rapidly expanding cloud marketplace. Much like the interconnected network economy that has evolved in the wireless world, the cloud is primed for the advancement of standards that will establish an open framework for cloud-to-cloud federation, benefiting both cloud providers and end users. This panel will discuss cloud standards in general, and address cloud federation standards underway in the IEEE and collaboration with the NIST Public Working Group on Cloud Federation. Panel Moderator: Stephen L. Diamond, VMware, Senior Director of Industry Standards, Industry Standards Office, Office of the CTO Steve is Senior Director of Industry Standards at VMware, a global leader in cloud computing infrastructure and digital workspace technology and part of Dell Technologies. Prior to VMware, he was Global Standards Officer and General Manager of the Industry Standards Office at EMC. He began his career at the Langley Porter Neuropsychiatric Institute at UCSF doing research in expert systems for EEG biomedical signal analysis. Steve is a life Senior Member of the IEEE, where he now chairs the Future Directions Committee. Before that, he twice served on the IEEE Board of Directors, was President of the IEEE Computer Society, and was the Editor-in-Chief of IEEE Micro Magazine. Steve received the IEEE Richard E. Merwin medal, the IEEE Computer Society Golden Core Award, and the IEEE 3rd Millenium Medal. Panelists Robert Bohn, NIST, Emerging Network Technologies Group Robert Bohn, of the Applied and Computational Sciences Division in NIST/ITL, serves as the Reference Architecture Lead for the NIST Cloud Computing Program. In this role, he works with industrial, academic and other government stakeholders to develop a high‐level vendor neutral reference architecture and taxonomy under the NIST Strategy for Developing a US Government Cloud Computing Technology Roadmap. This architecture will be used as a frame of reference to facilitate communication, illustrate and understand how clouds services and components fit together. Joel Fleck, CA Technologies, Principal, Industry Standards and Open Source, Office of the CTO Joel has served as a Board Director, executive, chief architect and lead contributor in numerous standards groups and activities including INCITS Cloud38, ISO/IEC JTC 1 SC 38, IEEE-SA, OASIS, OMG, WS-I and the TeleManagement Forum. Convenor of the JTC1 SC 38 Working Group on Cloud Interoperability and Portability, he led the team that developed the ISO International Standard for Cloud Interoperability and Portability. Currently, Joel serves as the Chair of the IEEE-SA P2303 Working Group on Adaptive Management for Cloud Computing. Joel graduated from the University of Michigan with a MS in Industrial and Operations Engineering specializing in large-scale system design and the University of Vermont with a BS in Computer Science with coordinate majors in Electrical Engineering and Environmental Engineering. He is Distinguished Fellow of the TeleManagement Forum. John Messina, National Institute of Standards and Technology (NIST), Supervisory Computer Scientist John Messina is a senior member of the Cloud Computing (CC) Project at the National Institute of Standards and Technology (NIST). He holds a M.S. in Computer Science and B.S. in Physics and has been working at NIST as computer scientist since 1998. He has a distinguished record of scientific accomplishments, such as publications and industry presentations, and has received several awards, including NIST's Edward Bennett Rosa Award and the US Department of Commerce's Bronze Medal. His focus has been on speeding up the adoption of Cloud Computing by United States Government agencies through the development of Cloud Computing vocabularies, standards, reference architectures, and guidance documents.
Friday, January 12 17:30 - 18:30
- 17:30 A Study on Class A Impulsive Noise Cancellation and Channel Estimation Under Rayleigh Fading Environment
- OFDM is used in many applications that require high-speed communication. However, in OFDM under impulsive noise, the reception characteristics are strongly deteriorated by impulsive noise because impulsive noise has short time duration and very large magnitude. Therefore, it is necessary to reduce the influence of impulsive noise. In the compensation of impulsive noise, the channel condition is estimated by using pilot symbols. In this paper, the reception characteristics are improved under impulsive noise by the proposed scheme using scattered pilot symbols.
- 17:50 A Novel Design of UHF RFID Passive Tag Antenna Targeting Smart Cards Limited Area
- The aim of this paper is to design an Ultra High Frequency (UHF) Radio Frequency Identification (RFID) passive tag antenna to fit in limited space of smart cards, such as bankcards, along with contactless payment facility and Europay, MasterCard and Visa (EMV) chip. In this paper, UHF tag antenna is designed using Monza R6 chip specifications from Impinj manufacturers while focusing on high performance and read ranges of up 12 meters with very small sized antenna dimensions of 17:9 47:30 mm2 . Performance is analysed in terms of gain, read ranges and power reflection coefficients. UHF tag antenna is designed using aluminium on Polyethylene Terephthalate (PET) substrate. The proposed design aims to meet objectives of low cost and small size while maintaining high performance.
- 18:10 Application Level Network Virtualization Using Selective Connection
- Since Internet-of-Things (IoT) devices transfer several types of data such as sensor values, images, and videos, multiple heterogeneous network interfaces are required to handle different types of data communications. To utilize the multiple network interfaces efficiently, the handover technique is important. However, the current network systems cannot transfer data during the handover operation between heterogenous network interfaces. In addition, the application programmer must implement the handover operation. In this paper, we propose a novel network virtualization technique called selective connection, with which an application-oblivious seamless switching between heterogeneous network interfaces is possible depending on the data type.
- 17:30 Prediction of the Time to Pregnancy by Kernel Density Estimation on Lifestyle Questionnaire
- We presents a personalized method to estimate the time to pregnancy (TTP) for a natural conception by using kernel density estimation and k-nearest neighbor. Using this method, we can provide suggestions and a TTP estimation to guide decisions on which methods and actioins are suitable for those willing to conceive. In addition, we developed a user-friendly GUI system that can show the probability of the TTP and provide feedback through tips without requiring user expertise.
- 17:50 Training Environment for Electric Powered Wheelchairs Using Teleoperation Through a Head Mounted Display
- Virtual wheelchair training environments have been implemented by different authors. However, they have not accurately represented the physical world behavior. Therefore this proposal consists of creating a training environment based on teleoperation, in which the user will remotely conduct a wheelchair receiving a real-time video feedback through an HMD.
- 18:10 Deep Learning for Automated Classification of Calcaneus Bone Fracture on CT Images
- Calcaneus, also called as heel bone, is the largest tarsal bone that forms the rear part of the foot. Cuboid bone articulates with its anterior and superior sides together with talus bone. Calcaneus is known to be the most fracture prone tarsal bone. Patient data can be stored in several kinds of imaging format, e.g. Computer Tomography (CT) data. It is a sequence of 2-D images that construct 3-D images. CT images contain a lot of medical information, such as fracture information in each slice of 2-D images that may not be accurate using visual inspection and need computer-assisted in solving this problem. This study proposed a new method to classify the fracture in calcaneus bone CT images. Convolutional Neural Network was applied in the classification step to see the better result between two methods. We classify into two classes (fracture and non-fracture) from three views that are coronal, transversal, and sagittal views.
- 17:30 S-CHIRP: Secure Communication for Heterogeneous IoTs with Round-Robin Protection
- This work introduces CHIRP - an algorithm for communication between ultra-portable heterogeneous IoT devices with a type of round-robin protection mechanism. This algorithm is presented both in its basic form as well as in a secured form in order to secure and maintain trust boundaries and communication within specific groups of heterogeneous devices. The specific target application scenarios includes resource constrained environments where a co-located swarm of devices (adversarial in mission or objective) is also present. CHIRP, and its secured version (S-CHIRP), enables complete peer-to-peer communication of a n-agent network of devices in as few as n-rounds. In addition to the n-round cycle length, the proposed communication mechanism has the following major properties: nodes communication is entirely decentralized, communication is resilient to the loss of nodes, and finally communication is resilient to the (re)-entry of nodes. Theoretical models show that even the secure implementation of this mechanism is capable of scaling to IoT swarms in the million device range with memory constraints in the <10 MB range.
- 17:45 Portable Tor Router: Easily Enabling Web Privacy for Consumers
- On-line privacy is of major public concern. Unfortunately, for the average consumer, there is no simple mechanism to browse the Internet privately on multiple devices. Most available Internet privacy mechanisms are either expensive, not readily available, untrusted, or simply provide trivial information masking. We propose that the simplest, most effective and inexpensive way of gaining privacy, without sacrificing unnecessary amounts of functionality and speed, is to mask the user's IP address while also encrypting all data. We hypothesized that the Tor protocol is aptly suited to address these needs. With this in mind we implemented a Tor router using a single board computer and the open-source Tor protocol code. We found that our proposed solution was able to meet five of our six goals soon after its implementation: cost effectiveness, immediacy of privacy, simplicity of use, ease of execution, and unimpaired functionality. Our final criterion of speed was sacrificed for greater privacy but it did not fall so low as to impair day-to-day functionality. With a total cost of roughly $100.00 USD and a speed cap of around 2 Megabits per second we were able to meet our goal of an affordable, convenient, and usable solution to increased on-line privacy for the average consumer.
- 18:00 Progressive and Secure Performance Unlocking for Digital Integrated Circuits
- This paper proposes a secure architecture to progressively unlock the performance of an integrated circuit based on the level of authorized access. A key comparison module authenticates the usage of the IC, uniquely identifies it and unlocks a specific mode of operation based on the authorized level of access. We present methods for tuning the performance at run-time based on the response from a key comparison module. IP owners can gain control over the activation, testing and usage of an IC by implementing the proposed authentication protocol. We have analyzed the proposed architecture on several ISCAS99 benchmark circuits and an image normalization application.
- 18:15 Secure Tunable-Precision Architecture for Image Processing Applications
- Numerous applications involving image and audio presentation, such as video games, can be made available to consumers with varying levels of precision based on the level of authorization. This paper presents a novel tunble-precision architecture for digital systems. In this architecture, computational precision can tuned across a large number of modes. A user can pay for and receive a key that unlocks one of the modes of precision which in turn enables a certain level of audio/image presentation and user experience. We present the secure tunable-precision architecture and an image processing application to demonstrate its usage.
- 17:30 The Future of Holographic Head Up Display
- This is an Invited Paper. The use of Holographic or Diffractive optics is on the rise, premium DSLR optics using this technology is an outstanding example of compact device. This technology applied to HUD is presented, using the windshield as a transparent holographic display, with the ability to present floating graphical object in a large field of view. Augmented Reality display will be possible, increasing considerably the User Experience and situational awareness. Additionally, the use of Holographic Optical Element reduce the size, weight and cost of the actual HUD box.
- 18:00 Prototype Gesture Recognition Interface for Vehicular Head-Up Display System
- The proliferation of infotainment systems in the contemporary vehicular environment offers an abundance of information which is typically not distilled or prioritised according to the driver's requirements and driving conditions. Additionally, this requires from the driver to operate multiple interfaces, resulting in an overwhelming cognitive load and a higher probability of collision. Our work presents a new multimodal Head-Up Display interface that aims to enhance human responses during a potential collision whilst moderate the plethora of incoming information through a simplified and manageable system. The latter is fully interactive and operated with the use of an embedded gesture recognition system. The proposed system was evaluated in contrast to traditional Head-Down Display interface by twenty users, with the use of a high-fidelity driving simulator. The results were in favour of the proposed system which reduces the collision propensity by 90% in a motorway environment.
(1) Tao Zhang, Corporate Strategy Group, Cisco Systems, email@example.com
(2) Bruce McMillin, Missouri University of Science and Technology, firstname.lastname@example.org
(3) Robert S. Fish, NETovations, LLC, email@example.com
(4) Tom Coughlin, Coughlin Associates, firstname.lastname@example.org
- 17:30 Fog Computing and Networking, What's Next?
- Clouds alone are becoming increasingly inadequate for supporting the emerging systems and applications, such as Internet of Things (IoT), 5G wireless systems, Big Data, edge analytics, and embedded Artificial Intelligence (AI). Fog computing and networking - or fog - has emerged to fill the gaps by bringing computing, networking, management, and control functions anywhere along the cloud-to-things continuum where these functions can best meet users' requirements. The immersive fogs can address many challenges that clouds cannot effectively address, such as enabling realtime edge analytics and local control, connecting and protecting the vast spectrum of resource-constrained devices, and overcoming network bandwidth and availability constraints. As fog will extend from the clouds all the way to the end users, it will have a profound impact on consumer communications. Wireless gateways in homes would become fog nodes that will serve not just as a networking gateway but also as a local computing and data storage server. A fog node embedded on a vehicle can make the vehicle a seamless part of the end-to-end services provided by the fog and the clouds to, for example, help update the software on the many microcomputers on the car and to allow the user to reconfigure the applications and user interface in the car. When someone gets in her car, her smartphone can become a fog node to provide services to her car. Consumer drones will rely on fog nodes on the ground and on other drones to help guide his flight regulate drone traffic, and ensure drones' safety and security. This new fog paradigm imposes both new opportunities and research challenges, calls for fundamental rethinking of computing and networking architectures, and can disrupt existing business models and reshape industry landscapes. The industries, from chipmakers to networking companies to software companies to application developers to IoT vertical technology providers and users, are devoting significant efforts to develop fog technologies. The OpenFog Consortium (OpenFog), consisting of industry movers and leading academic institutions are developing an open fog reference architecture and building a global ecosystem to accelerate market adoption of fog. The many profound research challenges in fog computing and networking are also drawing a booming interest in the academia. A growing number of universities and R&D organizations have launched fog-related initiatives. We have witnessed a breath of new workshops, panels, and journal special issues on fog computing and networking have over the past couple of years. Government agencies around the world are initiating new R&D programs on fog. On this plenary panel, industry and academia experts from different regions of the world will discuss their visions on the road ahead on fog computing and networking, focusing on challenges in fulfilling the full potential of fog computing and networking.
Friday, January 12 19:00 - 20:30
Welcome Reception Tonight Join your colleagues for our Welcome Reception this evening (Friday 12 January) at 7:00pm. Look for the ICCE signs in the Westgate Casino Sportsbook Area. Food, beverages, and networking will be provided.
Saturday, January 13
Saturday, January 13 8:30 - 9:00
Saturday, January 13 9:00 - 10:00
Keynote 3: Emerging NUI-based Methods for User Authentication,
Computer Science and Engineering,
New York University
- 9:00 Emerging NUI-based Methods for User Authentication
- As user demand and cost benefits of natural user interface (NUI) technologies are hastening their adoption, computing devices that come equipped with these interfaces are becoming ubiquitous. Consequently, authentication mechanisms on them are becoming an essential security component to enable a wider range of applications that need higher requirements of security as well as privacy. In this talk we will survey the landscape of "point-of-entry" user-device authentication mechanisms based on behavioral biometrics that require a natural user interaction using gestural or non-gestural interaction for access. This interaction includes 2-D touch gestures, 3-D gestures, voice, eye tracking, and braincomputer interaction. We will analyze their potential security and usability promises and issues, and discuss plausible solutions that could be pursued in future work.
Saturday, January 13 10:00 - 10:30
- 10:00 A Quadruple-polarized Reconfigurable Antenna for 915 MHz ISM Band Applications
- In this paper, an ultra-high-frequency (UHF) quadruple-polarized reconfigurable antenna for 915MHz industrial, scientific, and medical (ISM) band applications is proposed. The proposed antenna consists of a disk-shaped radiator, four port feeding network which integrates microstrip lines and a 90° hybrid coupler, and a SP4T RF switch for polarization control. By switching the operating modes on the feeding network, the quadruple-polarized reconfigurable antenna with LHCP, RHCP, VLP, and HLP can be obtained. Using a FR4 substrate, the proposed antenna was designed and implemented. The peak gain and antenna size at 915MHz of the proposed antenna are approximately 6.4 dBi(c) and 0.53λ0 (D) × 0.08λ0, respectively.
- 10:01 An LC Balun Integrated Microstrip TWA for Precise Near-Field Localization System in UHF RFID
- An LC balun integrated traveling wave antenna (TWA) for precise near-field localization systems is proposed. The proposed antenna fed by a grounded coplanar waveguide (CPW) line consists of a LC balun, 50 ohm loads, and differential-fed microstrip lines for uniform H-field distribution. To reduce a standing wave in the microstrip lines, 50 ohm resistors at both ends of the differential microstrip lines are inserted as loads. By differential currents along two microstrip line, the uniform magnetic field at the area between two lines is generated. The proposed compact TWA operates at the authorized universal UHF RFID band, and the measured impedance bandwidth based on the -15dB reflection coefficient is 0.84-0.95 GHz. The proposed compact TWA can be used for smart drawers or smart shelves which need precise localization.
- 10:03 One-side Automated Impedance Matching Scheme for Multi-transmitter Wireless Power Transmission
- This paper focuses on impedance matching issue for 2-to-1 (two transmitting (TX) coils to one receiving (RX) coil) coupled magnetic resonance (CMR) wireless power transmission system. It starts from discussing conventional 1-to-1 CMR system and its impedance matching issue. We then further studies the model of a 2-to-1 CMR system and compares the similarity and difference between them. Based on this, we propose a method to solve the impedance mismatch problem and an automation scheme to automate the impedance matching process. An experiment is conducted to validate the proposed method. And the result shows that with the proposed system can respond actively to the impedance mismatch caused by the spatial movement of receiver in a 2-to-1 CMR system. The power transmission efficiency are all enhanced for both transmitting coils.
- 10:04 Selective Mapping Based RF Watermark Technique for Robust Data Transmission in ATSC System
- In the Advanced Television Systems Committee (ATSC) digital television (DTV) systems, RF watermark technique was introduced to analyze interference among DTV transmitters for single frequency networks (SFN). Recently, there is an increasing interest in employing RF watermark technique for additional data transmission. This paper proposes a selective mapping based RF watermark technique for robust data transmission. Laboratory test results demonstrate that the proposed technique can achieve an enhanced data transmission capability with backward compatibility to legacy DTV receiver.
- 10:06 Analysis of Modulation Schemes for Bluetooth-LE Module for Internet-of-Things (IoT) Applications
- Bluetooth transceivers have been in the active area of recent research being the key component of physical layer in the Bluetooth technology. The low power consumption of Bluetooth low energy (LE) devices compared to the conventional Bluetooth devices has enhanced its importance in Internet-of-Things (IoT) applications. Therefore, Bluetooth low energy device based solution needs expansion in the IoT network infrastructure. The transceivers in the literature and modulation schemes are compared and summarized. Energy consumption of modulation schemes in Bluetooth communication are analyzed and compared using the model presented in this work. In this approach considering both circuit and signal power consumption, optimum modulation order for minimum energy consumption has been found using numerical calculation and relation between signal to noise ratio (SNR) and channel capacity. Battery life for IoT sensors using Bluetooth LE technology as a wireless link has been analyzed considering multiple transaction times for transmitters having different power consumption. MFSK and more bandwidth-efficient GFSK are identified as low energy solution for all smart devices.
- 10:07 A Study on LLR Calculation Scheme Under Mobile Reception of OFDM
- In the mobile reception of OFDM, ICI(Inter-Carrier Interference) is generated by Doppler-spread. To improve reception characteristics, the ICI canceller with complexity reduction scheme has been proposed. When actual systems are operated, error correction codes(ECCs) are usually used. In the calculation of LLR(Log-Likelihood Ratio), noise power is considered. However, the interference power is not considered in LLR calculation. In this paper, the scheme of LLR calculation considering interference power of ICI using Turbo code is proposed. The proposed scheme is evaluated by computer simulations. As the results, it is possible to improve the reception characteristics.
- 10:09 Toward Secure Packet Delivery in Future Internet Communications
- Information-centric networking (ICN) is new emerged Internet model that focuses on what is being exchanged rather than which network entities are exchanging information. This gives the control plane functions such as routing location the ability to be specified according to the content items. Bloom filter based forwarding is one candidate that is used in ICN for packet forwarding. The scheme has the advantage of being simple, stateless, and fast technique. Although, this forwarding scheme solve many problems the of today's Internet such as the growth of the routing table and the scalability issues, it is vulnerable to brute force attacks which are starting point to stop DDoS attacks. In this work, we design a novel source-routing and information delivery technique that keeps the simplicity of using Bloom filter based forwarding while being able to deter different attacks such as DDoS. The preliminary analysis of this proposal indicates that with the designed scheme, the forwarding function can detect and prevent malicious activities at early stage and with very high probability.
- 10:10 KKT-Conditions Based Resource Allocation Algorithm for DASH Streaming Service over LTE
- In this paper, we propose a DASH friendly resource allocation algorithm that enhances consumer's Quality of Experience(QoE) especially by reducing the re-buffering time ratio of video streaming and guarantees fairness among non-video traffic over LTE. The proposed method is based on Karush-Kuhn-Tucker(KKT)-Conditions and provides optimal solutions by considering buffer levels and past average throughputs of DASH and non-DASH consumers. Finally, we validate this by using NS-3(Network Simulator) to prove that our method shows better performance in the re-buffering time ratio and fairness among non-video traffic compared to existing resource allocation methods.
- 10:12 Evaluation of a Secure End-to-End Remote Control System for Smart Home Appliances
- In most existing remote-control services for smart home appliances, a controller outside the home network can be used to control these appliances in the home network using a remote-control support server installed on the Internet. However, if the service provider ends the service, a user cannot control these appliances from outside the home. To solve this problem, we have proposed a secure end-to-end remote-control system based on the "Network Traversal with Mobility," which can solve a NAT traversal problem and achieve end-to-end encrypted communication and IP mobility. In this study, we implemented a prototype of the proposed system and conducted a remote control experiment and performance evaluation for smart home appliances compliant with ECHONET Lite. Consequently, a user could directly access ECHONET Lite appliances inside the home from outside the home network with a low delay without any remote-control support server.
- 10:13 Development of Magnetic Resonant Wireless Power Transfer System Robust to Position Gap
- Recently, as a convenient charging method for small IoT sensors, a magnetic resonant wireless power transfer (WPT) system is attractive. In WPT systems, robustness of the efficiency due to the displacement of the receiver and transmitter is very important. In this paper, we propose a method that reduces the efficiency decrease even if a position is displaced by using a Helmholtz coil structure. Comparison with general single coil WPT system is discussed in the evaluation about the efficiency.
- 10:15 Precoding Design for Two-Way MIMO Relaying with Antenna Selection
- In the paper, joint source-and-relay optimization is proposed for two-way multiple-input multiple-output (MIMO) amplify-and-forward (AF) multi-relay systems with using a layered relay-and-antenna selection (LRAS) mechanism under power constraints in correlated channels. This minimum mean-squared error (MMSE) based optimization results in a highly nonlinear objective function such that an iterative algorithm is utilized to jointly optimize the source, relay, and receive matrices under power constraints. Simulation results show that the joint architecture of transceiver/relay precoders in conjunction with the LRAS strategy is capable of accomplishing superior performance in terms of MSE and bit-error-rate (BER) compared with other existing transceiver design algorithms.
- 10:16 Standard Cell Camouflage Method to Counter Silicon Reverse Engineering
- Payment and identification IDs using smart cards technology are vulnerable to a physical attack by reverse engineering the netlist of an embedded integrated circuit. An attacker gains access to a netlist by de-packaging and delayering a chip and processing its micrograph images. We proposed a low performance impact and novel method to obfuscate gates by using a current digital design flow with a layout standard cell generator to obfuscate non-critical paths. Results indicate that at cell and block level, a 15% timing and power overhead is incurred in applied worst-performance benchmarks. After applying the method to a set of benchmark circuits, results shown a 40% average obfuscation, or not detected gates, of the resultant netlist.
- 10:18 Fast-LSTM Acoustic Model for Distant Speech Recognition
- The distant-talking automatic speech recognition (ASR) currently becomes an important task in a speech recognition area. Traditionally, hybrid Gaussian Mixture Model-Hidden Markov Model (GMM-HMM) approach are used for ASR. This paper will discuss some deep neural network (DNN) techniques for acoustic modeling, as well as lattice rescoring techniques for ASR. The proposed Fast-long short-term memory neural network (Fast-LSTM) acoustic model combines the time delay neural network (TDNN) and LSTM network to reduce the training time of the standard LSTM acoustic model.
- 10:19 Systematic Network Coding Based Reliable Real-time Multimedia Streaming System
- In this paper, we propose an implementation of a real-time multimedia streaming system that can simultaneously support a large number of clients. The proposed solution is based on systematic network coding that does not require feedback channels for acknowledgment or retransmission while it is resilient to packet erasures. Our experiment includes simultaneous full HD video transmission to 20 clients with 4 different types of mobile devices. The measured results confirm that the proposed system is the most robust against the packet losses compared to two conventional real-time streaming solutions.
- 10:21 Pixel-based Fast CU Depth Decision Algorithm with Edge Strength for HEVC
- In this paper, an optimal coding unit (CU) depth decision algorithms are proposed to efficiently reduce the computational complexity for inter coding in High Efficiency Video Coding (HEVC). The optimal CU depth is calculated by the characteristics analysis of the texture using sobel-filter. Moreover, the combinational method of CU depth based on texture analysis is implemented for the optimal CU depth decision. Simulation results show that the proposed algorithm will induce PSNR loss 0.060dB with only 2.49% bitrate increase, compared to HM10.0. Because our proposed algorithm can decide the optimal CU depth before encoding start, our pixel-based fast CU decision algorithm greatly contribute to the reduction of hardware area and the power consumption. Additionally, by using our proposed algorithm, the highly parallel processing can be implemented.
- 10:22 An Extension Nonnegative Matrix Factorization for Occluded Image Recognition
- This paper addresses the challenge of recognizing face and facial expression under occlusion situations. We have introduced an extension of nonnegative matrix factorization called angel and graph constrained nonnegative matrix factorization (AGNMF). The proposed model is developed in term of incorporating the minimizing angel of basic cone and preserving the geometrical structure of the projective data. The experimental results in the context of recognition occluded images demonstrate that the technique of enforcing constraints on both basic and encoding matrices works well and the AGNMF method shows superior performance to other conventional approaches
- 10:24 Variable Resolution View Multiplexing for Mobile 3D Display
- In this paper, we present a novel view multiplexing technique that enhances perceptual resolution in mobile autostereoscopic 3D displays. An autostereoscopic display shows two or more views that have binocular disparity at the same time, so it allows the viewer to perceive motion parallax at the scene. Generally, the display divides the display panel spatially and assigns R, G, B color sub-pixels of multiple view images at the proper position on the display. It is called sub-pixel multiplexing or interweaving. Each viewpoint image resolution is the same as the resolution of a target display to show a 3D image without loss of quality. However, the amount of data is too huge to deal with the images for practical application. The proposed method significantly reduces the amount of data needed by minimizing the degradation of visual quality.
- 10:25 Coordinate-Based Direction-of-Arrival Estimation Method Using Distributed Microphones
- In this paper, we propose a new coordinate-based di-rection-of-arrival (DOA) estimation method, which is applied to several distributed security cameras that have a single microphone each. The proposed method tries to specify the location coordinate of each sound source at regular intervals. To this end, the time dif-ferences between multi-channel microphones are first estimated by applying the generalized cross-correlation with phase trans-form (GCC-PHAT) algorithm to the signals recorded in each pair of microphones. Then, a support vector machine (SVM) is used to classify whether each sound source is located at a closed area con-structed by the microphone array. Next, the location of the sound source is estimated as the point at which there is a minimum error between the estimated time difference from the recorded signal to each pair of microphones and the time difference at the specified coordinates using GCC-PHAT. A performance evaluation shows that the classification accuracy of the SVM is 98% and the average distance error is within 10cm.
- 10:27 Design and Performance Evaluation of a Localization System to Locate Unwanted Drones by Using Wireless Signals
- The use of consumer drones has increased greatly in the last few years. It is important in many situations to be able to determine the location of unwanted drones for the sake of privacy, security, and safety. In this paper, a localization system is designed to determine the location of an unwanted drone in an outdoor line of sight (LOS) situation by using its wireless signal transmitted for communication with the controller. Two algorithms using the received signal strength (RSS) of the unwanted drone, the recursive least squares (RLS) algorithm and an improved maximum likelihood estimation (MLE), are implemented. Experiments are conducted in the open areas of the university campus to obtain a 3D RF empirical propagation model and this channel model is used in the algorithms. The performance of the two algorithms are compared with each other and with the Cramer-Rao Lower Bound (CRLB). By analyzing the results, we conclude that the improved MLE algorithm performs better than the RLS in our localization system. We also discuss the challenges that we will face in the future to improve our system.
- 10:28 Interoperability and Reliability of Wearable Healthcare Devices for Emergency Care
- Wearable consumer healthcare devices are becoming increasingly popular over recent years with a range of non-invasive sensors that continually track a user's health. These consumer wearables differ from medical devices mainly due to the fact that they do not need to undergo the stringent FDA approval process. One of the main advantages of these wearables is to provide important information about the user's state of health in the event of an emergency. Given that there is currently no standards for consumer healthcare devices, interoperability for information retrieval by paramedics in emergency situations become the utmost important aspect in ensuring interconnection between these devices.
Saturday, January 13 10:30 - 12:00
- 10:30 Conversion of LDR Image to HDR-like Image Through High-Level Synthesis Tool for FPGA Implementation
- High dynamic range (HDR) of an image will increase the visibility of the image. Due to the reflection on any scene, the image may lose its visibility in some specific area. In the image, we call it highlight. We cannot reveal the original color and information in the highlight area. By removing this highlight, we can increase the visibility of the image that expands the dynamic range of image. In this paper, we convert our algorithm by high-level synthesis (HLS) tool to the synthesizable hardware language. Our target is to show that our algorithm for HDR-like image can be implemented on FPGA. We also compare our HLS results with simulation result.
- 10:48 Early Mode Selection of High Resolution for HEVC Base on Bits-Mapping
- Modern video use higher resolutions and frame rates to offer better visual quality. However, with super high resolution, the computation complexity of the encoder is increased. In this work, we proposed a low complexity algorithm based on the relationship between the header bits and CU partitions. Firstly, a frame is divided into eight regions, then a map tendency of the bits distribution is created. Next, based on the map, the complexity reduction algorithm is proposed. Finally, the proposed algorithm is evaluated by HM15.0. The simulation results show that the proposed algorithm can achieve over 50% computation complexity reduction.
- 11:06 Multiple Sports Player Tracking System Based on Graph Optimization Using Low-Cost Cameras
- Multiplayer tracking is a key technology for strategic analysis in team sports. Since existing solutions use high-resolution cameras and are also expensive to install, they are primarily for professionals and there are almost no solutions that consumers can easily obtain. We focused on K-Shortest Paths which performs well with low-end cameras located at a tripod-level height, where occasional detection failure may occur. We implemented a prototype that can generate visual tracking results based on the algorithm, investigated factors that influences tracking performance and computational time using multi-camera video sequences, and estimated computation time for tracking multiplayer in several team sports.
- 11:24 Hardware-oriented Low Complexity Motion Estimation for HEVC
- In this paper, we proposed an adaptive search range selection algorithm to reduce complexity of the motion estimation. In previous work, the method which determines the search range using motion vector variance was proposed. However, the selected search range is not efficient enough. The proposed algorithm can efficiently reduce the redundant search range using the distribution of the surrounding motion vectors. In addition, the proposal performs the hierarchical motion estimation when the search range is still too wide. The simulation results show that the proposed algorithm achieved averagely 80.9% complexity reduction with 0.9% BD-rate increasing, compared with the previous work.
- 11:42 Depth Extraction Using Depth of Field Imaging with Tilted Retroreflective Structure
- We propose the method of depth extraction using the tilted retroreflective structure which consists of a beam splitter and micro corner cube array. A camera takes the reconstructed object image by retroreflector with narrow depth of field. The image contains the focused object image and focused area within tilted retroreflector. The area indicates the position of the object image. Therefore, we can estimate the object position with simple calculation such as the lens equation. The proposed method is useful for real-time depth extraction without active light source. The experiment is performed to compare with the theoretical calculations, of which results confirm the feasibility of the proposed method.
- 10:30 Depth Map Estimation with Unknown Fixed Pattern Projection
- The simple two-plane calibration is proposed to estimate depth with fixed NIR random dot projection and camera system. The depth refinement algorithm with an RGB image is also proposed. Experimental results demonstrate that the simple calibration is good enough to estimate the depth map. It is also shown that the depth map can be refined by the RGB camera image.
- 10:48 A New Approach to Motion Judder Cancellation: Time Stamp Model Using Instant Frame Rate
- Motion compensated frame interpolation is a key technology to remove motion blur and motion judder in displays. We propose a new approach to motion judder cancellation using time stamp model using instant frame rate. The proposed method exactly identifies the original frame rate of input frames and effectively reduces motion judder by avoiding abrupt changes in detection result. Also, the proposed method is robust against digital noise and bad edit. We can reduce motion judder noticeably by combining the proposed method with any high quality motion compensated frame interpolation techniques.
- 11:06 Sound Source Separation for Plural Passenger Speech Recognition in Smart Mobility System
- This paper proposes a novel sound source separation (SSS) for a multi-path automatic speech recognition (ASR) system corresponding to simultaneous utterances in a smart car. The proposed method significantly reduces insertion errors in the ASR, which is caused by interfering speeches of passenger, and makes it possible to respectively recognize each desired speech present in a target direction with high accuracy even though plural passengers utter.
- 11:24 Driving Method for Suppressing Image Distortion in Tiled OLED Displays
- Tiled displays made by arranging organic light-emitting diode (OLED) displays are one of candidate for large and ultra-high-definition OLED displays. In an OLED display unit, however, there is a scanning delay between the first and last horizontal lines because the image is created line by line. This delay causes image distortion at the joints of tiles in the display because the first and last lines of adjacent tiles are showing different frames. Here, we propose a novel driving method to suppress this image distortion in tiled OLED displays.
- 11:42 Robust Remote Heart Rate Estimation in Car Driving Environment
- In this paper, we propose a novel framework for estimating driver's heart rate remotely under driving condition. First, region of interest is selected by landmark points derived from discriminative response map fitting. Feature signal is then extracted for heart rate estimation and refined by dynamic standard deviation check to eliminate the noisy segments. Finally, heart rate is estimated by power spectral analysis using temporal filtered signal. We test our framework on video database captured from real-world car driving condition. Experimental results validate that the proposed method achieves better results compared with the most prominent conventional methods.
Closing Remarks: H. Okumura (Toshiba, Japan)
- 10:30 Wide Field of View Optical Combiner for Augmented Reality Head-up Displays
- Wide field of view combiners for head-up displays are reviewed and our originally developed combiner using Fresnel reflector that acts both as a concave reflector for reflected light and a flat plate for transmitted light has been introduced. Main issue of ghost image derived from diffraction and refraction is eliminated by Fresnel pattern with variable pitch, sandwiched by the materials with the same refraction index.
- 10:48 Motorcycle HUD Design of Presenting Information for Navigation System
- This paper introduces the motorcycle head-up display design of presenting information for navigation system. In particular, a human-centered design approach was adopted to design as a human-machine interface. Therefore, a motorcycle head-up display prototype was developed to perform tests to design information presentation for navigation system. Virtual reality environment was utilized as a test environment, with an operational motorcycle simulator equipped with a head-up display. For experimental subjects, motorcycle riders with license participated riding the operational simulator. The experimental results answered several questions of presenting information design using head-up display for motorcycle.
- 11:06 LcAR - Low Cost Augmented Reality for the Automotive Industry
- Currently, AR (Augmented Reality) technologies for the automotive industry are available mostly for high-end luxury vehicles in developed countries where the nature of the solution is such that it is very costly to install and the repair cost is also high. However, for developing countries, there is a huge and ever-increasing population that drives at night in low-light conditions where car accidents are almost usually fatal. For developed countries too, the fatalities due to animal crossings etc is high. This paper presents a mechanism to drastically improve the visibility in low-light conditions using Augmented Reality where the real-time video is analyzed, and augmented with the highlights indicating obstacles if any. Our system uses an IR camera with suitable range. This video is then processed to identify the obstacles via machine learning frameworks, classified appropriately and augmented by highlighting the obstacles in the driver's path either visually or audio feedback or both. Our system can build upon previous knowledge to enable constant learning and improvement in the detection accuracy.
- 11:24 Phase Hologram 3D Head Mounted Displays Without Zero Order Diffraction
- We have proposed phase hologram 3D head mounted display using optical system eliminating zero-order diffraction. We clarified that zero-order diffraction light can be eliminated without reducing the field of view of the display by the spatial filter that blocks only straight component of light.
- 11:42 Focus-Free Head-Mounted Display Using Pinhole and Retroreflector Film
- We propose a focus-free retinal projection system, which is composed of a display device without coherent light, a pinhole, and a retroreflector film. By virtue of the retroreflector film as an image conjugator, a pinhole conjugate is formed as a small exit pupil at the center of eye pupil, which enables focus-free retinal imaging regardless of focal length of eye lens. This focus-free property of the proposed system addresses accommodation-vergence mismatch of three-dimensional augmented reality display system. We performed experiments to show the focus-free property of our proposed system.
- 10:30 Deep Learning in Low-Power Stereo Vision Accelerator for Automotive
- Various forms of Convolutional Neural Network (CNN) architectures are used as Deep Learning (DL) tools for learning the similarity measure on video patches in order to run the stereo matching algorithm - the most computationally intensive stage of the pipeline for the stereo vision function used in designing an autonomous car. We propose a hybrid system implementation of the algorithm for real-time, low-power and high-temperature environment. The accelerator part of our system is a programmable many-core system with a Map- Reduce Architecture. The paper describes and evaluates the proposed accelerator for different versions of the stereo matching algorithm.
- 11:00 End-to-End Pedestrian Collision Warning System Based on a Convolutional Neural Network with Semantic Segmentation
- Traditional pedestrian collision warning systems sometimes raise alarms even when there is no danger (e.g., when all pedestrians are walking on the sidewalk). These false alarms can make it difficult for drivers to concentrate on their driving. In this paper, we propose a novel framework for an end-to-end pedestrian collision warning system based on a convolutional neural network. Semantic segmentation information is used to train the convolutional neural network and two loss functions, such as cross entropy and Euclidean losses, are minimized. Finally, we demonstrate the effectiveness of our method in reducing false alarms and increasing warning accuracy compared to a traditional histogram of oriented gradients (HoG)-based system.
- 11:30 Detecting a Pothole Using Deep Convolutional Neural Network Models for an Adaptive Shock Observing in a Vehicle Driving
- A pothole is a one of the greatest threat to vehicle drives. It causes an accident by sudden steering of the vehicle wheel, forcing an enormous stress on a vehicle tire or making a hard turning in a vehicle by late detection. It is crucial to find where a pothole is on the pavement. As the number of pavement increases, detecting a pothole becomes a great challenge in a modern society. Several methods suggest detecting potholes using sensors. However, these methods require an installation on the vehicle in order to collect data of the pavement. Meanwhile, other methods are using smartphone sensors to reduce a cost of deployment and get an advantage of sensitive sensors without a complex installation on the vehicle. For this reason, a method using a smartphone camera with the artificial neural network becomes a way in detecting a pothole on a pavement. In this paper, we investigate the performance in detecting potholes with an image classification method based on the deep convolutional neural network models.
(1) Introduction - from Brain Computer Interface, Brain Data Bank to Brain Communication, N. Nan Chu, CWLab International, USA, and National Central University, Taiwan.
(2) Using EEG to quantify the enhancement of cognitive control deficits in older adults following video game training, Joaquin A. Anguera, Director of Clinical Program, Dept. of Neurology and Psychiatry, UC-San Francisco, USA.
(3) The 1st Brain Data Bank Challenge in St. Petersburg, Russia, Konstantin Glasman, St. Petersburg State University of Film and Television, IEEE CESoc TV Chair, and Yuri Shelepin, Pavlov Institute of Physiology, Russian Academy of Sciences, Russia.
(4) A package for prediction post-stimulus reactions based on pre-stimulus EEG signal, Anna Stetsenko, Evgeny Blagoveshchensky, Olga Vakhrameeva, Petr Vasiliev, Alexey Harauzov, Pavlov Institute of Physiology, RAS, Russia.
(5) Brain Data Bank Competition in Boston - Lessons Learned and Future Directions, Seth Elkin-Frankston, Charles River Analytics, USA and Wasim Malik, Harvard University & MIT, USA.
- 10:30 Brain Data Bank (BDB) Visualization, Analytics and Beyond
- IEEE Brain Initiative has conducted international Challenges and Competitions, to explore creative means for investigation of open brain EEG datasets. The goals are to: • raise awareness about Brain Signal System Technology • assess usability of available brain signal datasets • explore Big Data Analytics, Artificial Intelligence, Deep Learning for user-friendly brain signal databank. • facilitate brain signal data formatting standardization The winning projects are highlighted. Many ingenious ideas were demonstrated throughout these events. For example: Lego mobile feeder, brain wrestling, brain dataset performance contrast, and open source tool creation for novice computer users to integrate brain EEG data collected under various environments.
Saturday, January 13 12:00 - 13:30
- 12:00 Deep Learning Networks in CE
- I could probably offer a half-way answer to "what is the next challenge for CE Imaging?". Reality is that I don't know for sure (there are many), but I know how to solve it: neural networks. The real question becomes how to enable low cost NN implementations and deployments in CE devices without the fear of losing years of work invested in training and optimizing the networks designed to solve specific problems (sound enhancement, better imaging, understanding of the surroundings, easier cooking, better coffee, etc..). This talk will detail the challenges and industry proven solutions for computer vision and computational imaging using a hybrid traditional imaging and deep learning approach. IP protection against intellectual theft once the solutions are deployed is also briefly discussed.
Saturday, January 13 13:30 - 14:45
(1) Ram K. Krishnamurthy, Senior Principal Engineer, Intel Corporation, USA (2) Travis Humble, Director, Quantum Computing Institute at Oak Ridge National Laboratory, USA (3) Sen-ching "Samson" Cheung, Professor, University of Kentucky, USA (4) James Lyke, Program Manager, Air Force Research Laboratory's Space Systems Branch, USA (5) Saraju P. Mohanty, Professor, University of North Texas, USA (6) Matthew Casto, Chief of the Air Force Research Laboratory, Trusted and Assured Electronics Branch at Wright-Patterson Air Force Base in Dayton, OH, USA
- 13:30 Energy and Cybersecurity Constraints on Consumer Electronics
- With the growth of Internet-of-Things (IoT) enabled consumer electronic devices, the potential threat vectors for malicious cyber-attacks are rapidly expanding. As an example of cyber-attack, software vulnerabilities could be exploited to remotely take control of safety-critical systems in the vehicle. These cyberattacks are threat to the reliability and safety of the consumer electronic devices, consumer's personal information and piracy or cloning of intellectual property. As the IoT paradigm emerges, there are challenging requirements to design area-efficient, energy-efficient and secure systems. Further, due to novel computing paradigms such as quantum computing there is a threat that the fundamental public-key cryptography tools could be broken. Considering these challenges, the panel would provide their perspective towards the energy and cybersecurity constraints on consumer electronics.
Saturday, January 13 14:45 - 15:30
- 14:45 Age Category Estimation Using Matching Convolutional Neural Network
- This paper proposes an age category estimation method using matching convolutional neural network (CNN). The proposed method compares an input facial image with target facial images using matching CNN and classifies the input facial image into one of three cases: younger than, similar to, or older than target images. Using a pair of the input facial image and the facial image in each age category, the proposed method dramatically improves the accuracy of the age category estimation.
- 14:47 Performance Acceleration of Neural Networks on Mobile Embedded Systems
- In this paper, we describe applying techniques for accelerating neural network algorithms on mobile embedded systems. MNIST handwriting artificial neural networks that identify numerical images are accelerated and compared using OpenCL, OpenMP, NEON, and other performance optimization techniques.
- 14:49 Time Delay Convolutional Neural Network for Acoustic Scene Classification
- This paper proposes a novel neural network framework that can be applied to commercial smart devices with microphones to recognize acoustic contextual information. Our approach takes into consideration the fact that an acoustic signal has more local connectivity on the time axis than the frequency axis. Experimental results show that the proposed method outperforms two conventional approaches, which are Gaussian Mixture Models (GMMs) and Multi-Layer Perceptron (MLP), by 8.6% and 7.8% respectively in overall accuracy.
- 14:52 CNN-based Approach for Visual Quality Improvement on HEVC
- Deep learning based on Convolutional Neural Network (CNN) is a very hot issue for various recognition problems in this paper, we proposed a scheme to improve the visual quality of the video coding standard. We apply a CNN model to high efficiency video coding (HEVC) encoding and suggest combined scheme using a CNN model. Through experiment, we verify that the proposed scheme achieves up to 0.24 dB of the quality improvement on HM 16.10 reference software.
- 14:54 Locally Adaptive Contrast Enhancement Using Convolutional Neural Network
- Contrast enhancement plays a crucial role in image processing. Because of external factors(environment, backlight), obtained images have low contrast which can reduce performance of image processing systems, such as in intelligent traffic analysis, visual surveillance, and consumer electronics. Particularly, low contrast images reduce visibility from the perspective of digital device users. Conventional image processing methods are proposed to solve these problems but the increase in overall brightness value causes some side effects such as over enhancement. In this paper, a method is proposed which detects low contrast region with CNN and to prevent some side effects we use chromatic contrast weight.
- 14:57 Extended Faster R-CNN for Long Distance Human Detection: Finding Pedestrians in UAV Images
- Recently, using consumer Unmanned Aerial Vehicles(UAV) for aerial photography has became a trend. However, the images captured from the UAV raise a challenge to the existing pedestrian detection algorithms, because the humans in the image are too blur and too low-resolution resulted from the long distance between the UAV and pedestrians. The problem of detecting long distance humans in an image has always been overlooked, so even the performance of the state-of-the-art detection algorithms are not satisfactory when used on UAV pedestrian detection. In this paper, we extend Faster R-CNN algorithm by proposing an improved Region Proposal Network(RPN) and utilizing object context information to improve the detection performance. The experimental results show that the extended algorithm improves the performance of detecting pedestrians captured by UAV.
- 14:59 Acoustic Scene Classification Using Convolutional Neural Networks and Multi-Scale Multi-Feature Extraction
- Audio scenes are often composed of a variety of sound events from different sources. Their content exhibits wide variations in both frequency and time domain. Convolutional neural networks (CNNs) provide an effective way to extract spatial information of multidimensional data such as image, audio, and video; they have the ability to learn hierarchical representation from time-frequency features of audio signals. In this paper, we develop a convolutional neural network and employ a multi-scale multi feature extraction methods for acoustic scene classification. We conduct experiments on the TUT Acoustic Scenes 2016 dataset. Experimental results show that the use of multi-scale multi feature extraction methods improves significantly the performance of the system. Our proposed approach obtains a high accuracy of 85.9% that outperforms the baseline approach by a large margin of 8.7%.
- 15:01 CNN-based LDR-to-HDR Conversion System
- We propose a system to convert low-dynamic-range images to high-dynamic-range (HDR) images. The system is based on a convolutional neural network (CNN). We use the CNN to distinguish light sources from non-light sources, and use that information to identify regions that should be brightened. Then the contrast is stretched, and the brightness of the region is further enhanced to obtain an HDR image.
- 15:04 Design of a Smart Greenhouse System Based on MAPE-K and ISO/IEC-11179
- A smart greenhouse monitors the internal and external context information of the greenhouse in real time and keeps the internal environment in an optimal condition for the growth of crops. In this paper, we propose a smart greenhouse system based on MAPE-K that can automatically control the greenhouse. All information of the proposed system is stored and managed through an ontology knowledge repository based on ISO-IEC 11179 (metadata registry, MDR). Furthermore, we design a low cost smart greenhouse by interoperating the smart devices.
- 15:06 Image Processing-based Pothole Detecting System for Driving Environment
- The pothole is one of the significant factors for the motorist. It can cause a car accident, decrease of car lifetime and the decrease of the motorist concentration. There are three kinds of the pothole detecting methods, such as using a 3-axis acceleration sensor, a camera sensor and, a laser sensor. All of three have the cons and pros. Our proposed system is based on an image processing method with the camera sensor. Detecting the potholes based on the potholes characteristic. The characteristics of the potholes are the dark region, the round shape and the rugged texture. The proposed algorithm is based on parallel processing with superpixel, wavelet energy field, and differential. It has high accuracy in Korean highway video
- 15:09 Methodology for Improving Detection Speed of Pedestrians in Autonomous Vehicle by Image Class Classification
- We propose a pedestrian detection method to minimize the amount of computation for classifying and candidate region detection in autonomous vehicles. The minimization of the computational complexity is a crucial factor for commercial products with a limited computational power. In conventional pedestrian detection methods, the number of candidate regions is 300 to 2,000 even if there is no pedestrian in an image. Therefore, the unnecessary computation is significant to classify each falsely decided candidate region. Moreover, it leads to false detection. In this paper, we propose a new methodology for solving this problem, and show through experiments that the processing speed can be improved by the proposed methodology.
- 15:11 Adaptive Color Balance Algorithm for Users' Conveniences
- In this paper, we propose adaptive color balance algorithm using auto-parameter. The algorithm analyzes histogram of each RGB channel and adjusts normalization coefficients, considering color distortion and contrast. Experiment results showed that the proposed method could reduce color distortion and increase contrast. This is useful for mobile phone camera and digital camera, where a user can freely acquire images.
- 15:13 WATT:a Novel Web-Based Toolkit to Generate WebAssembly-based Libraries and Applications
- 15:16 Personal Characteristics Expression by Changing Forms of the Eyelid for Deformed Face Robot
- Telexistence robot is felt a strong presence of its operator by personal characteristics expression. We propose a deformed face robot that can expresses personal characteristics by changing forms of the eyelid. We verify effects on individual identification by personal characteristics expression using the eyelid of the proposed robot. From the result, three types form (slanting eyes, normal eyes and drooping eyes) is recognition rate over 70% when opening-and-closing-degree are 0% to 30%. The proposed robot show that it can express personal characteristics using the eyelid. The proposed robot realizes a mental care application by giving a strong presence to the families who are in a remote place as an altered ego of operator's own.
- 15:18 Making Online Content Viral Through Text Analysis
- Enormous number of content are being published on the World Wide Web everyday in different forms such as blogs, news articles, movie reviews, product reviews, etc. Although huge number of unstructured content are being published it is questionable and unpredictable, how far a content reaches and engages a large number of expected audience. May it be blogs, news articles, movie reviews, product reviews, etc. The goal of our study is to develop a system that makes a post or textual content to go viral on the web when it is published. To make a post go viral on WWW, paper proposes a two step approach. First is to improve the content of the post by incorporating emotions and sentiments by deriving rules. Next is to take a post directly to its potential audience by measuring its popularity and give suggestions as how to improve the post in order to make it viral, based on the derived rules. The suggestions will be to replace certain words in the post. Paper focuses on social media posts of potential viewers of a post and analyzes what kinds of decisions they may take in future. Based on these analytics our system recommends relevant posts to viewers. We evaluate our approach using social media posts related to movie domain. Our system extracted text related to user's expectations to watch a certain kind of movie in the near future and recommend with the reviews about the movies he may likely to watch.
- 15:21 Efficient Data Cluster Management Scheme for Qcow2-based Virtual Disk in Home Cloud Server
- Nowadays, home cloud servers have adopted virtualization technologies to fully utilize their own computing resources. In this paper, we introduce a novel data cluster management scheme for qcow2-based virtual disk to improve I/O performance and storage space efficiency of the home cloud server. Our scheme aims to mitigate the sync amplification problem of qcow2 format, caused by allocating data clusters in a grow-only manner. To achieve the goal, we try to reuse the allocated data clusters that are no longer in use. Our experimental results show that our scheme improves I/O performance and reduces the number of sync operations, by up to 52.7% and 59.8%, respectively, compared with the conventional scheme.
- 15:23 A Framework for View Progressive Coding of Light Field
- This paper presents a framework enabling progressive access to light field views aiming at reducing the network load and storage capacity issues commonly related to light field applications. The proposed architecture is based on grouping light field views into fully configurable arrangements followed by standard image or video coding. A consumer device on the bases of the light field rendering application requirements sends a request to the light field server with the set of views, the coding format to be applied and the coding parameters. The light field server processes the request and sends the encoded bitstream to the consumer device.
- 15:25 Hardware and Software Implementation of A New Algorithm on Photoacoustic Medical Imaging
- In this paper, we review a new image reconstruction algorithm proposed by our research team and comment on the hardware and software implementation for that algorithm. We use TI embedded AFE board and xilinx FPGA combined with camera interface of freescale i.MX6 MPU for compactness of implementation and develop application program for preview image of raw B-scan image based on SDK firmware IDE. We use four segment of frame buffer for capturing raw B-scan image and store it to these four segments of frame buffer alternatively and move one segment of frame buffer to the LCD controller to remove and fix the flickering problem in LCD display. We comment on the GPU and CUDA programming for implementing of a new image reconstruction algorithm
- 15:28 Sparseness Subspace and Large Basic Cone Constrained Nonnegative Matrix Factorization with Kullback-Leibler Divergence for Data Representation
- Nonnegative matrix factorization (NMF) and its extensions have been intensively studied in computer vision. In this work, a new constrained NMF with Kullback-Leibler (KL) divergence model is developed for data representation. It is called basic cone and sparseness constrained nonnegative matrix factorization with Kullback-Leibler divergence (conespaNMF_KL). They achieve sparseness from a large simplicial cone constraint on basic matrix and sparse regularize on the extracted features. Experiments on two popular CK+ and JAFFE datasets for facial expression recognition (FER) reveal that the resulting subspace of the proposed algorithm achieves better enriched representation in comparison with the prevalent NMF methods
Saturday, January 13 15:30 - 17:00
- 15:30 A Novel Video Watermarking Approach Based on Implicit Distortions
- In order to protect videos from copyright infringement, a watermarking approach is proposed based on implicit distortions generated by a video encoder, rather than artificial distortions used in the state-of-the-art. These distortions are imperceptible and robust against video manipulations.
- 15:48 Light Field Compression by Superpixel Based Filtering and Pseudo-Temporal Reordering
- In this paper we have addressed the topic of an evolutionary integration of light fields into standard image/video processing chains by pre-processing light fields with superpixel-based and structurally adaptive Gaussian pre-filters and circular pseudo-temporal sequencing to feed them into an HEVC-codec with low-delay predictive coding configuration. We could show significant bit rate reductions of up to 27% compared to pseudo-temporal sequencing without pre-processing. The paper includes experimental results showing that not only the perceived visual quality, but also the cornucopia of post-processing options is preserved.
- 16:06 Self-interference Digital Holographic Camera Using Geometrical Phase Lens
- A self-interference digital holographic camera with geometrical phase lens used as a passive-type common-path polarization selective wavefront modulator is proposed. A prototype is devised and demonstration of self-interference is carried out. Since the geometrical phase lens is light weight, thin, transparent, and cost effective, the application toward the consumer level holographic camera capture system is expected.
- 16:24 Experimental Prototype of SD Memory Card Recordable 8K/60P Camcorder
- We have developed an experimental prototype of 8K camcorder that can record a compressed 8K/60p video signal on four SDHC/SDXC memory cards. This prototype system is composed of camera and recorder units. The camera unit acquires an 8K/60p video signal using a 33-megapixel CMOS image sensor with Super 35 optical format. The recorder unit encodes the uncompressed 8K/60p video to four square-divided 4K/60p video signals using an AVC/H.264 encoder, and stores them individually on four SDHC/SDXC memory cards in MOV format. The data rate of uncompressed 8K/60p video is 48 Gbps and that of the encoded 8K/60p video is 600 Mbps (4 × 150 Mbps); thus, the compression ratio is 80:1. The acquired video can be edited by commercial editing software that supports the 8K format. Most of the compressed 8K video shot by the camcorder has no visual artifacts and this is applicable to general 8K content production with simple editing.
- 16:42 Hybrid Autofocus System by Using a Combination of the Sensor-Based Phase-Difference Detection and Focus-Aid Signal
- We proposed a hybrid autofocus system which combines the PDAF with focus-aid (FA) signal, facilitating precise focusing with a low-resolution viewfinder. The experimental results showed that the FA signal information secured an accurate AF, and the PDAF improved the AF acquisition time without unwanted lens hunting phenomenon.
- 15:30 Visible-Light Wearable Eye Gaze Tracking by Gradients-Based Eye Center Location and Head Movement Compensation with IMU
- In this paper, a visible-light based wearable eye gaze tracking technique is developed for outdoor and indoor applications. By using the gradients based iris center location and head movement compensation with the inertial measurement unit (IMU), the proposed eye tracker works effectively. In our tests, the average estimated errors of iris centers are smaller than 6 pixels. For the indoor case at testing mode, the average horizontal and vertical accuracies of gaze tracking are 2.4 and 2.53 degrees, respectively. For the outdoor case at testing mode, the average gaze tracking accuracies are less than 2.2 degrees. The proposed core function executes up to 234 frames/sec by a PC at 3.4GHz working frequency.
- 15:48 Color Navigation by Qualitative Attributes for Fashion Recommendation
- This paper proposes a novel method to navigate a color palette using attributes recognized from speech input. Our target application is a fashion recommender system for mobile e-commerce. Starting with a selected color, a user can request to show items of a different color by qualitative attributes ( e.g. 'a little cuter'). These attributes are mapped to a query vector within the Lab color space in order to select the next color. The system distinguishes 85 attributes, each with three different possible magnitudes. This color navigation by speech was demonstrated in a mobile fashion recommender system. The proposed model is validated in a user study with 196 subjects.
- 16:06 Data Flow Synchronization of a Real-Time Fusion System to Detect and Recognize Smart TV Gestures
- This paper presents the data flow synchronization aspects of running a fusion system on a modern laptop in real-time. The fusion system uses two differing modality sensors of a Kinect depth camera and a wearable inertial sensor to detect and recognize a number of actions of interest from continuous action streams. This system is utilized to detect and recognize smart TV gestures when they are performed in a random and continuous manner among various actions of non-interest. It is shown that the processing times associated with the components of the developed fusion system lead to the real-time operation of the system on modern laptops.
- 16:24 Wearable Brain Computer Interface (BCI) to Assist Communication in the Intensive Care Unit (ICU)
- Lack of communication in hospitals can be a significant factor in causing distress for both patients and doctors, particularly in the Intensive Care Unit (ICU). The use of computers in this context, Brain Computer Interface (BCI) systems in particular, can facilitate communication on demand. BCI systems enable ICU patients to communicate using the electrical activity of their brains. For this purpose, we designed and developed a BCI system comprised of an Android tablet that allows patients to look at the screen to call the nurse, or ask for what they need solely using patients Electroencephalogram (EEG) recorded using a wireless wearable BCI. The flickering icons are rendered on the tablets screen, each conveying a specific message using an openGL-based paradigm stimuli. Meanwhile, the EEG signals are recorded wirelessly via Bluetooth (BT) in real-time. Subsequently, our designed and developed signal processing module interpret the EEG dynamics and identifies the flickering icon that the subject is focusing on. Our experimental results on 10 subjects demonstrated 98.7% average SSVEP identification accuracy.
- 16:42 Random Jitter and ISI Removal from High Speed Links Using Kalman Filtering
- In this paper, a new and simple approach is presented to remove random jitter and intersymbol interference (ISI) effects from NRZ (Non-Return-to-Zero) signals. The method presented uses a Kalman filter to predict and remove the random jitter noise and Intersymbol Interference (ISI). Random jitter is modeled as a Gaussian distribution, with jitter effects on both rise and fall times of the signal. The Kalman filter adapts based on the provided model and predicts the exact position of the rise and fall times of the clock signal and a look-up table, combined with the Kalman filter, predicts the exact level of the signal. Random jitter and ISI simulations in MATLAB are presented and the results obtained are highly promising. The method provides a new approach to remove jitter from high-speed links.
- 15:30 Mobile Lock Screen UI Design for Children to Enhance Positive Emotions Using Face Recognition Technology
- In this study, we designed an effective user interface (UIs) for lock screens using facial expression recognition technology and two interface methods to develop an application that induces positive emotional flow in children. There is not enough study considering practical use of lock screen with special characteristics for different types of user such as children. After that, in order to understand its effectiveness, the whole process on the methodology timeline consisted of two phases. As a result of the survey, children preferred a lot of tasks on the lock screen whereas adults who were the control group preferred a lower task load and were more likely to use the face recognition lock screen than the children. Based on the characteristics of the derived children, we expect to develop effective applications by integrating the results of the research and specifying the functions.
- 15:45 Markerless Augmented Reality Technology for Real-space Basketball Simulation
- For years simulation has been used in sport field for various purposes such as entertaining, training and others. One of the technologies that expected to contribute in sport simulation is immersive technology such as Augmented Reality (AR) and Virtual Reality (VR). This paper introduced the exploration of Markerless Mobile AR application for Basketball Simulation. In this project, the author showed that the virtual content that blend naturally with the real space could present a simulation project that close to the real field experience and it is critical point for sport simulation.
- 16:00 Technology Trend in 38 Live Entertainment Shows on the Las Vegas Strip
- About 46 million tourists, almost the same number as the population of Spain, visit Las Vegas every year. The Las Vegas Strip is a special area of 4.2 miles (6.8 kilometers) in length on Las Vegas Boulevard, where world class major hotels and entertainment facilities are located. This paper introduces the characteristics of live entertainment shows that are performed on the Las Vegas Strip. There are 34 hotels on the Strip identified and 15 of those hotels provide a total of 38 live shows, most are presented up to five times per week - early shows start at 1PM and late shows start at 10PM. Around 20 live shows out of the 38 are children friendly and target family visitors. Regarding the main technology being used in the 38 live shows, 55% of it is associated with lighting and video such as LED walls and strobe lights, 32% is associated with sound, and 13% is associated with motion control equipped in the live shows. Interestingly, no shows use virtual reality, augmented reality, or wearable technology.
- 16:15 Unmanned Aerial and Ground Vehicle (UAV-UGV) System Prototype for Civil Infrastructure Missions
- This paper develops notional technical design re- quirements for a team of autonomous air and ground vehicles for civil infrastructure inspection. Such a heterogeneous robot team can produce augmented maps of regions of interest with wider coverage and reduced uncertainty compared to mapping done by either platform alone because of the ability to exploit drastically different vantage points. In cases where visual scanning provides insufficient data, the aerial vehicle can deliver a ground vehicle equipped with additional sensors for nondestructive testing. State-of-the-art implementations support the validity of such a system, and preliminary testing suggests future work.
- 16:30 Multi-VR System of Asia World Unesco Buddhist Heritage
- Multi-VR System of Asia World Unesco Buddhist Heritage
- 16:45 Personal Tracked Vehicle Autonomy Platform
- There are available small form factor powered personal transportation devices for intracity travel, but most for only paved surfaces widely unintended to tow a load. We demonstrate the capability of a low-cost tracked vehicle for development of compact automated all-terrain transportation. This vehicle shows viability in heavy-duty applications.
- 15:30 End-to-End Deep Learning for Autonomous Navigation of Mobile Robot
- This paper proposes an end-to-end method for training convolutional neural networks for autonomous navigation of a mobile robot. Traditional approach for robot navigation consists of three steps. The first step is extracting visual features from the scene using the camera input. The second step is to figure out the current position by using a classifier on the extracted visual features. The last step is making a rule for moving the direction manually or training a model to handle the direction. In contrast to the traditional multi-step method, the proposed visuo-motor navigation system can directly output the linear and angular velocities of the robot from an input image in a single step. The trained model gives wheel velocities for navigation as outputs in real-time making it possible to be implanted on mobile robots such as robotic vacuum cleaner. The experimental results show an average linear velocity error of 2.2 cm/s and average angular velocity error of 3.03 degree/s. The robot deployed with the proposed model can navigate in a real-world environment by only using the camera without relying on any other sensors such as LiDAR, Radar, IR, GPS, IMU.
- 15:52 Light-weight Visual Place Recognition Using Convolutional Neural Network for Mobile Robots
- Place recognition is one of the essential components for mobile robot navigation. In this work, we present a light-weight convolutional neural network (CNN) approach for visual place recognition. Our proposed approach specifically targets embedded systems. To reduce the computational complexity in the network, we design a fully convolutional network architecture with fewer layers and filters. The proposed network directly learns a vector space where its distance corresponds to place similarity by metric learning. For effective training, we adopt triplet embedding with image dataset captured at various viewpoints. Such approach allows the network to embed scenes directly without any further training process on robots. The experimental results show that the proposed method significantly outperforms conventional algorithms including other CNN-based approaches in terms of accuracy and computational time.
- 16:15 The Application of Deep Learning on Depth from Multi-Array Camera
- Consumer-level multi-array cameras are a key enabling technology for next generation smartphones imaging systems. The present paper aims to analyze the accuracy of the depth estimation while using different camera combinations in a multi-array camera. This is done by providing a framework of deep neural networks to determine depth map from a sequence of images captured by a multi-array camera. Capturing depth information enables users to perform a range of post-capture edits such as refocusing, and creating a 3D model of any scene. Thus it is essential to calculate an accurate depth map while using the minimum computational resources
- 16:37 Deep Learning for Hand Segmentation in Complex Backgrounds
- This paper presents a Deep Learning segmentation approach for hand segmentation in gray level images with cluttered backgrounds where standard techniques cannot be used. Two networks were trained with a database of hand images derived from widely used palmprint image databases, Hong Kong Polytechnic University (HKPU) and Chinese Academy of Science (CASIA). The image dataset is augmented with complex patterns and used to train and test the Neural Networks, providing promising results.
Invited Talk: Veena Misra, North Carolina State University
- 15:30 Add Health Consumer Wearables Ancillary Study
- Background As of August 2017, there are over 424 wearable devices available to consumers from 266 vendors. Concurrently, researchers have begun considering the value of participant-generated health data to complement traditional data collection. Rapid proliferation of these technologies has left a considerable void in what is known about the validity, reliability, feasibility and long-term benefit associated with self-tracking. Despite concerns, researchers are deploying these technologies in support of public health interventions and clinical trials. The Fitabase Research Library has inventoried 435 studies published since 2012; a search of ClinicalTrials.gov using keyword "Fitbit" reveals 175 registered trials. It is important to understand more about the distribution of wearable device adoption, including the representitiveness of such populations, the characteristics of adopters, and patterns of tracking behaviors over time. The National Longitudinal Study of Adolescent to Adult Health (Add Health) is a longitudinal study of a nationally representative sample, recruited as adolescents, in grades 7-12 across the United States during the 1994-95 school year. Add Health is re-interviewing cohort members throughout 2016-2018 to collect social, environmental, behavioral, and biological data with which to track the emergence of chronic disease as the cohort enters their fourth decade of life. This study will survey 13,000 individuals drawn from the Add Health cohort to capture data on smartphone and consumer wearable device ownership. From eligible respondents identified in the survey, we will obtain historical measures of physical activity captured using personally owned wearable devices. Objectives Aim 1: Determine rates of smartphone and consumer wearable device adoption within the Add Health cohort. Aim 2: Invite cohort members identified in Aim 1 to make a one-time donation of historical physical activity data collected using personally owned wearable devices. Aim 3: Prepare a harmonized and standardized consumer wearable dataset for linkage with other existing Add Health data (e.g., social, economic, physical, contextual measures). Aim 4: Conduct preliminary analysis of wearables data, including device types, duration and consistency of tracking, and physical activity outcomes. Methods We have introduced new survey items, drawn from the Pew Research Center's questionnaires on the same topics for comparison. RTI's Participant-Generated Health Data platform is used to electronically consent and extract data from consumer wearable device vendor web services. Analysis will focus on preparation of descriptive statistics regarding technology adoption in various subgroups, (e.g., high or low adopters), characteristics of those participants, and potential for collecting other types of biometric data from the Add Health cohort based on the devices members own. We expect the data collected in this ancillary study to be made available as a public-use dataset in early 2018. Results Data collection was initiated in June 2017 and is expected to continue until the end of the calendar year. As of August 2017, 263 respondents completed the survey items and 29 respondents (11%) agreed to donate data from their wearable devices. No analyses had been completed at the time of submission. Summary statistics of the survey responses and descriptive statistics of wearable datasets will be prepared for ICEE2017.
- 15:45 Wearable Sensors for Air Force Applications
- Within the Materials and Manufacturing Directorate of the Air Force Research Lab (AFRL), we are addressing the specific materials and processing challenges that are preventing wearable electronics from meeting the mission-specific needs of the United States Air Force (USAF). Advancements in wearable electronics are expected to impact a wide range of missions, including aircraft pilots and crew, special operators, aeromedical evacuation personnel, and various analyst teams. Specific projects to be presented include the research and development of a smart flight suit for pilot health monitoring and a confined space monitoring system for improving the safety of aircraft maintenance personnel as well as failure analysis efforts for next-generation wearable electronics. We are working to address these challenges through in-house research into new materials, innovative packaging schemes, and new approaches to conformal and integrated electronics. Additionally, we lead both NBMC and NextFlex - America's Flexible Hybrid Electronics Manufacturing Institute to grow and foster the flexible hybrid electronic ecosystem using industry- and university-led projects. The presentation will address the unique opportunities that universities, government labs, and industry have to engage with our teams to grow their understanding of and capability to create wearable electronics that are relevant to the USAF Mission.
- 16:00 Wearable Sensors for Brain EEG Signal Oriented Applications
- Various sensors are compared for development toward non-invasive and wearable brain signal collection, processing, correlation, and applications. Since the last decade, consumer-grade brain-computer interface (BCI) headsets and brain EEG datasets have become available to trigger digital intelligence development in conjunction with augmented and virtual reality capabilities. Such neurotechnology advancement has found intriguing applications in social networking, entertainment and autonomous driving9. Combining stimulus from vision, facial and other physiological signals have also been trialed to demonstrate the effectiveness of EEG sensors on wearables.
- 16:15 i-Helmet: An Intelligent Motorcycle Helmet for Rear Big Truck/Bus Intimation and Collision Avoidance
- In this paper, we propose an intelligent motorcycle helmet, called i-Helmet, which integrates IR sensors with an image sensor and adopts the image recognition methodology to recognize rear big vehicles. Two detection modes (day/night) are designed for the purpose of the image recognition accuracy. Experimental results showed that the proposed i-Helmet can successfully be achieved image recognition of the license plate for rear big truck/bus. The recognition accuracy rate achieves about 70% (night) to 75% (day). Therefore, the proposed i-Helmet can real-time provide related intimations for avoiding rear big truck/bus collisions.
Saturday, January 13 17:00 - 18:30
- 17:00 Edge-aware Facial Skin Beautification Based on Skin Tone Probability
- In this paper, we propose a novel facial skin beautification method which can smooth out skin textures in a natural manner. More specifically, the smoothing process of the proposed method is mainly focus on uniform skin textures while preserving the edges and boundaries of facial features or non-skin parts, thereby preventing from generating the so-called blurry facial images. For this, the proposed method utilizes a pixel-wise blurry mask which combines both edgeness and skin tone probability of each pixel. In various experiments, the proposed method shows very promising performance in terms of both the processing speed and the beautification results.
- 17:18 Improvement of H.265/HEVC Encoding for 8K UHDTV by CU Size Expansion and Inter/Intra Prediction Mode Selection
- For 8K UHDTV video, which has high spatial resolution, CU size has large influence to encoding efficiency. In addition, a moving object with random motion has also a large impact on encoding efficiency of inter prediction. In this paper, we therefore propose a method to improve H.265/HEVC encoding efficiency by CU size expansion and inter/intra prediction mode selection. Moreover, we evaluated encoding efficiency and complexity by chancing max CU sizes and inter/intra prediction modes. From these evaluation results, the proposed method achieved high efficiency and low computational cost.
- 17:36 Monocular Color-IR Imaging System Applicable for Various Light Environments
- In this paper, we propose a monocular color-IR (MCIR) imaging system that can capture color and near infrared (NIR) light simultaneously under various lighting conditions. Our system captures visible and infrared light by using a single 2 × 2 color filter array (CFA) image sensor with an optical notch filter and merges both signals into a single image. To precisely reproduce color the same as color cameras, we considered separating RGB signals from a captured image, adjusting the white balance for various lighting color temperatures, and adjusting the chrominance saturation in accordance with various RGB and IR lighting ratios.
- 17:54 A Study on Color-space Conversion Method Considering Color Information Restoration
- In this paper, we consider a color space conversion method from BT.709 used for current HDTV (High Definition Television) broadcast to BT.2020 that will be used for UHDTV (Ultra HDTV) broadcast, with lost color information restoration. Our method anisotropically diffuses BT.709 chromaticities considering the direction to the original chromaticities in BT.2020 color space. With this method and BT.709 images, we obtained BT.2020 images that had chromaticities out of BT.709 color gamut.
- 18:12 Efficient Intra Prediction Based on Adaptive Downsampling Signal for Parallel HEVC Encoding
- Intra coding utilizes neighbouring reference pixels to construct the prediction samples and reduce spatial redundancy. However, the traditional intra prediction method in High Efficiency Video Coding (HEVC) is a real hindrance for parallel hardware implementation and making further improvement on coding efficiency. In this paper, an efficient parallel scheme is proposed for HEVC intra coding. Downsampling signal is utilized to generate intra prediction instead of neighbouring pixels and remove the data dependency in intra encoding for coding tree unit (CTU) structure. Meanwhile, a fast training method is designed to supply downsampling signal adaptively. Experimental results show that proposed fast parallelized scheme achieves 4.17% bit saving on average, with reducing computational complexity by 27.26%.
- 17:00 Accurate Positioning of Bicycles for Improved Safety
- Cyclists are not well protected in accidents and there are few active safety systems available for bicycles. We have evaluated the use of inexpensive RTK-SN receivers together with dead reckoning for accurate positioning of bicycles to enable active safety functions such as collision warnings.
- 17:22 Estimating Driver's Readiness by Understanding Driving Posture
- In this paper, we propose a novel Driver Monitoring System (DMS) that estimates whether drivers are able to control vehicles. This system detects driver's inability state to control vehicles by learning normal driving posture. We implemented this DMS on a real vehicle for evaluations. 10 participants drove the vehicle on three test courses and we checked our DMS performance in real time. The evaluation results show that this novel DMS has potential performance for practical use.
- 17:45 Simulation Framework for Improved UI/UX of AR-HUD Display
- The AR-HUD (Augmented Reality-Head Up Display) overlays the ADAS (Autonomous Driver Assistant System) information to real world objects on the windshield unlike the normal HUD. This projected ADAS information on the windshield usually occurs the irregularity between the objects and the ADAS information because of the difference between driver and AR-HUD's view plane. To overcome this problem, we present a simulation framework for the improved AR-HUD in ADAS by means of a homographic registration with inference outputs from a region based deep learning model. In order to build this simulation framework, we set up a simulation test bed to mimic on-road driving environment in a darkroom, and use the inference model based on region-based fully convolutional network to obtain on-road ADAS information. And then we apply the homographic registration method to minimize the irregularity between object and ADAS information in terms of driver's perception. We tested the proposed simulation framework with real world driving recordings, and it showed better display results for improved UI/UX with inference outputs from region-based deep learning model.
- 18:07 Driver's Gaze Zone Estimation by Transfer Learning
- Estimating driver's gaze zone has very important role to support advanced driver assistant system (ADAS). The gaze estimation can monitor the driver focus and indirectly control the user interface / user experience (UI/UX) on windshield using augmented reality-head up display (AR-HUD). However, to train gaze zone estimator as classification task, someone pays huge costs to gather a large amount of annotated dataset. To reduce the labor work, we used transfer-learning method using pre-trained CNN model to project the gaze estimation task by regression on mobile devices that has large and reliable dataset into new classification task to overcome lack of annotated dataset for gaze zone estimation. We tested the proposed method to our own building simulation test bed. The result shown in validation accuracy around 99.01 % and test accuracy with unseen driver around 60.25 % for estimating 10 gaze zones in-vehicle.
- 17:00 Enhancing Iris Authentication on Handheld Devices Using Deep Learning Derived Segmentation Techniques
- In this paper, the increasing use of biometric authentication on handheld devices is considered and the importance of accurate iris segmentation in implementing an embedded authentication workflow based on iris authentication is explained. A deep learning scheme is then developed and an appropriate augmentation method is presented to solve the problem of the iris segmentation task in handheld devices. Initial comparisons with publicly available iris segmentation algorithms show significant performance improvements, particularly on challenging image datasets designed to mimic the image quality obtained from a handheld device.
- 17:22 Learning Data Augmentation for Consumer Devices and Services
- Transferring the success of deep learning models to consumer electronic devices requires the construction of deep learning models that are small enough to fit on resource constrained hardware. Since embedded and mobile devices lack the resources in terms of power consumption requirements, processing speed, and available memory of the latest GPU technology, it is desirable to create neural networks that are significantly smaller without sacrificing accuracy. A new technique for data augmentation called "Smart Augmentation" has recently been introduced that has been experimentally shown to be effective at this task. In this paper, we show how Smart Augmentation can be used to train models that are significantly smaller than their equivalently performant counterparts, and thus more viable for deployment on consumer devices.
- 17:45 Efficient Light Harvesting for Accurate Neural Classification of Human Activities
- The energy autonomy extension of wearable devices is an ever increasing user need and it could be achieved by inexpensive energy harvesting from the broadly available solar and artificial light. However efficient conversion, relevant storage and utilization must be carefully implemented if the device supports power-hungry applications such as Artificial Intelligence for human activity classification based on Artificial Neural Networks. In this paper, a whole hardware and software system implementation is presented, which is able to achieve system autonomy extension and at the same time high classification accuracy. Quantitative and qualitative results are shown under real working conditions.
- 18:07 Efficient SIMD Implementation of Binarized Convolutional Neural Network
- This paper presents the efficient implementation of the binarized convolutional neural network (CNN). The conventional CNN model is modified by including the binarization processes so that each of the parameters and the feature elements can be represented by a single bit, and trained maintaining the analysis performance. Multiple parameters and feature elements are packed into a single word and the inner-product operations are implemented efficiently by employing the bit-wise XNOR followed by the bit-count operations. By such SIMD optimization, the proposed implementation achieves a $9.5\times$ speed-up in terms of the inference time for AlexNet, when compared to the straightforward implementation. In addition, its memory footprint is $9.5\%$ of that required in the straightforward implementation.
- 17:00 A Comparison of Amazon Web Services and Microsoft Azure Cloud Platforms for High Performance Computing
- Advances in commercial cloud computing motivate the necessity to continually evaluate the cloud's performance on a variety of applications. This work looks at compute oriented instances from Amazon Web Services and Microsoft Azure cloud platforms and evaluates them with several high performance computing benchmarks, including HPCC and HPCG. These benchmarks illustrate that the most cost competitive solution depends on the application to be run.
- 17:18 CMD: A Convincing Mechanism for MITM Detection in SDN
- There are many security problems in network environment. Among them, man-in-the-middle (MITM) attack, as a typical network attack, greatly threatens the communication of network and damages victims. Until now, solutions to reduce the MITM attacks still have many limits. In this paper, we proposed a proactive detection mechanism CMD to detect MITM attacks in SDN network. The topology and connection characteristics of network traffic are utilized by CMD, without the analysis of packet contents. Through the theoretical analysis and experimental verification, CMD shows good time and accuracy performance.
- 17:36 Light-weight Accountable Privacy Preserving (LAPP) Protocol to Determine Dishonest Role of TPA in Cloud Auditing
- Cloud computing is surfacing as the next disruptive utility paradigm. It provides large storage capabilities, the development environment for application developers through virtual machines. It is also the home of software and databases that are accessible, on-demand. As security is the main constraint holding companies to fully engage into cloud, third party auditors (TPA) are becoming more and more common in cloud computing implementations. Of course, involving auditors comes with its own issues such as trust and processing overhead. To achieve productive auditing, we need to (1) accomplish efficient auditing without requesting the data location, nor introducing processing overhead to the cloud client; (2) avoid involving new security vulnerabilities during the auditing process. In this paper, we introduce a novel method allowing to detect a dishonest TPA: the Light-weight Accountable Privacy Preserving (LAPP) Protocol. This protocol determines the malicious behavior of the TPA. To validate the effectiveness of the proposed protocol, simulation experiments have been conducted, using the GreenCloud simulator. Based on our simulation results, we confirm that our proposed protocol provides better outcomes as compared to the other known contending protocols in terms of reliability, communication cost, and the auditing time per task by a fraction of the invalid responses.
- 17:54 A Containerized Media Cloud for Video Transcoding Service
- Videos transmitted on the Internet need multiple versions due to the variety of needs from users. Video providers need to transcode an original video into different formats, with different bitrates and resolutions to satisfy those needs. To meet such demands, we design and implement a novel lightweight containerized media cloud for video transcoding service in this paper. We show the high efficiency, robustness, scalability and portability of our system.
- 18:12 Efficient Cloud Tracing: From Very High Level to Very Low Level
- With the increase of cloud infrastructure complexity, the origin of service deterioration is difficult to detect because issues may occur at the different layer of the system. We propose a multi-layer tracing approach to gather all the relevant information needed for a full workflow analysis. The idea is to collect trace events from all the cloud nodes to follow users' requests from the cloud interface to their execution on the hardware. Our approach involves tracing OpenStack's interfaces, the virtualization layer, and the host kernel space to perform analysis and show abnormal tasks and the main causes of latency or failures in the system. Experimental results about virtual machines live migration confirm that we are able to analyse services efficiency by locating platforms' weakest links.
(1) Invited Talk, Dave Wang, Striiv Inc. (2) Invited Talk, Ryan Kraudal, Valencell Inc. (3) Invited Talk, Natalie Wisniewski, Profusa Inc.
- 17:00 A Band-Pass IR Light Photodetector for Wearable Intelligent Glasses in a Drowsiness-Fatigue-Detection System
- This paper proposes a band-pass infrared (IR) light photodetector, which is applied to our developed wearable intelligent glasses in a drowsiness-fatigue-detection (DFD) system. The proposed band-pass IR photodetector is designed to detect IR light in the wavelength range from about 810 to 890nm. The benefits of the proposed band-pass infrared photodetector are to provide higher SNR, decrease the ambient environmental light image to a minimum, and effectively enhance detection accuracy. As a result, the recognition algorithm is easier to be implemented and mounted on the light-weight version wearable intelligent glasses with high detection accurate rate.
- 17:15 Smart Belt for Recommending Food Items Based on the Stomach Dynamics
- Majority of the people treat food habits, the intake quantity and the body status as independent things leading to health issues. Although devices are available in the market today that tell the calorific value and other attributes of the food and body status of the person such as hunger separately, there is no mechanism to tell how long a certain food stuff takes for digestion and how much of it should be consumed. It is person specific and changes dynamically. In this paper, an ultrasound sensor is embedded in the waist belt to estimate these parameters
Sunday, January 14
Sunday, January 14 8:30 - 9:00
Sunday, January 14 9:00 - 10:00
Keynote 5: Internet of Things Security: Are We Paranoid Enough?
Swarup Bhunia, Electrical and Computer Engineering,
University of Florida
- 9:00 Internet of Things Security: Are We Paranoid Enough?
- Security has become a critical design challenge for modern electronic hardware. With the emergence of the Internet of Things (IoT) regime that promises exciting new applications from smart cities to connected autonomous vehicles, security has come to the forefront of the system design process. Recent discoveries and reports on numerous security attacks on microchips and circuits violate the well-regarded concept of hardware trust anchors. It has prompted system designers to develop wide array of design-for-security and test/validation solutions to achieve high security assurance for electronic hardware. At the same time, emerging security issues and countermeasures have also led to interesting interplay between security, verification, and interoperability. Verification of hardware for security and trust at different levels of abstraction is rapidly becoming an integral part of the system design flow. The global economic trend that promotes outsourcing of design and fabrication process to untrusted facilities coupled with the prevalent practice of system on chip design using untrusted 3rd party IPs, has given rise to the critical need of trust verification of IPs, system-on-chip design, and fabricated chips. The talk will also cover spectrum of security challenges for IoTs and describe emerging solutions in creating secure trustworthy hardware that can enable IoT security for the mass.
Sunday, January 14 10:00 - 10:30
Sunday, January 14 10:30 - 12:00
- 10:30 RRPhish：Anti-Phishing via Mining Brand Resources Request
- Although in recent years a variety of anti-phishing studies have been carried out, phishing fraud has become increasingly rampant. Especially with the popularity of electronic banking and mobile payment, phishing attacks have become more profitable. In this context, exploring efficient and practical anti-phishing technology is particularly necessary and urgent. In this paper, by analyzing the resources (CSS, JS, and image files) request characteristics of phishing sites, we propose a new anti-phishing method -- RRPhish, which is applicable to both client and server scenarios. In the client scenario, RRPhish as an enhanced blacklist technology, can detect not only phishes in blacklist, but also emerging phishes. The client-side experiments demonstrate the effectiveness of RRPhish, which is an effective complement to existing anti-phishing methods.
- 10:48 Support for ECHONET-based Smart Home Environments in the universAAL Ecosystem
- With the advent of information and communication technology, many Ambient Assisted Living solutions are being proposed to increase the quality of life of elderly people and reduce health and social care costs. Among these AAL solutions, universAAL seems to be the most promising platform for easy and economical development of AAL services. However, in its current state, the platform is incompatible with smart home systems which are based on the ECHONET standard. This paper presents the bridging between the universAAL and ECHONET standards through a technical point of view and thereby enables AAL services for ECHONET-based smart home environments.
- 11:06 Study of Personal Customized Life Assisting System Consisting of Modular Devices
- We developed a personal customized life assisting system consisting of modular devices to give useful information for us when we go out and communicate with people such as family and friends. Personal customized modular connections became possible by using Pogo pins in the connection part of the modules. Experimental results confirmed that our proposed system is useful for outgoing support services.
- 11:24 Heterogeneous Data Based Danger Detection for Public Safety
- In recent days, public safety service attracts attentions because of increasing crime rates. In this paper, heterogeneous data based danger detection (HDDD) mechanism is proposed for reducing crime rates. The HDDD mechanism is a mechanism for detecting dangerous situations (e.g., homicide and violence) based on heterogeneous data that are data gathered from multiple devices such as closed circuit television (CCTV) cameras, smartphones, and wearable devices. The HDDD enables immediate detection of dangerous situations and then reduces crime rates by responding to the dangerous situations.
- 11:42 Fuzzy Based Prediction Schema Framework for IoT Based Indoor Environmental Monitoring
- In recent times, rapid introduction of the Internet of Things (IoT) services in smart environment domain has happened. This has introduced heterogeneous device and data types that lack proper mechanism for joint execution of tasks from an application perspective. In this paper, we propose an integrated indoor environmental monitoring system implemented with a novel IoT framework. It is enabled via fuzzy-based rule schema to resolve the heterogeneity of environmental data in smart home environment. The novel aspect of the framework is that it is developed with a modular repository together with decision-making modules. Our experiments prove this approach to be a viable IoT solution for smart home indoor environment.
- 10:30 Stereoscopic realtime 360-degree video stitching
- We present a system for real-time capturing and stitching high-resolution stereoscopic 360-degree video, which is suitable for streaming and viewing by a variety of existing VR devices. Our novel stitching algorithm uses depth proxy to resolve parallax problem and render omnidirectional stereo (ODS) video. Our 2-step camera calibration algorithm for multiple lenses allows finding precise intrinsic and extrinsic parameters, which in turn, allows doing accurate depth estimation and robust and consistent stitching. Both depth estimation and stitching algorithm are optimized for parallel computation on GPU, so we are able to stitch VR video at realtime.
- 10:45 Content Adaptive Tiling Method Based on User Access Preference for Streaming Panoramic Video
- Tiled streaming has been proposed for delivering ultra-high resolution videos such as zoomable online lectures or panoramas. However, not much work has been done on finding the best tiling method for streaming panoramic video. This paper proposes an effective tiling algorithm for tiled streaming by using both video content and user access preference history. Experimental results show that the proposed tiling method can save up to 32.4% and 69.8% of average streamed bitrate compared to conventional uniform tiling scheme and simply streaming the entire panorama respectively on equi-rectangular panoramic video.
- 11:00 Evaluating Accessibility Features Designed for Virtual Reality Context
- The current work aims to present the evaluation of the efficiency of the Accessibility features on a Virtual Reality (VR) environment. The objective was to test an open source solution named Gear VRF accessibility, which provides a framework to be used by developers, and has the purpose of adapting Zoom, Inverted Colors, Auto-reading (Screen Reader) and Caption features in a VR environment. Interested to cover all the aspects related to the performance and satisfaction of the accessibility features hypotheses were created to be tested in a controlled environment. Regarding the fact that the solution was embedded into the Samsung Gear VR, all the participants were introduced to this device before testing the accessibility features. The performance and the level of satisfaction observed while the user executed the tasks were combined with users statements and resulted in a set of positive and negative feedbacks to improve the interaction and usage of the solution tested.
- 11:15 Functional and Computational Requirements for Mobile Augmented Reality Applications
- Functional and computational requirements have been dealt in designing mobile AR applications, considering theoretical studies, practical usability and technical constraints. Specifically, knowledge representation and graphical overlay rendering were mainly focused with the corresponding scenarios.
- 11:30 Panoramic Video Delivery Based on Laplace Compensation and Sphere-Markov Probability Model
- Virtual Reality (VR) is on the rise nowadays. To save bandwidth and ensure high quality when supplying VR videos, only the tiles overlapped with field of view (FOV) are transmitted in high quality while other tiles in low quality or even not transmitted. However, if a user's viewpoint moves fast, low-quality content will always appear in FOV because of round-trip time (RTT) delay. We propose methods based on Laplace compensation and Sphere-Markov probability to increase high-quality area in FOV. And a strategy is proposed that the former should be exploited over low-RTT networks while the latter over high-RTT networks.
- 11:45 Enabling Autonomous Navigation in a Commercial Off-The-Shelf Toy Robot for Robotic Gaming
- Commercial off-the-shelf (COTS) robots are becoming ever more ubiquitous. Among the most common applications there are toy robots. In order to keep their cost down, these robots are usually equipped with the minimum set of sensors necessary for their basic functioning. Specifically, to add entertainment value, they often feature a video camera and, in most of the games developed so far, players are basically tele-operating them through a smartphone app. The present paper aims to show how to provide existing consumer-grade robots with new capabilities to transform them in more appealing gaming companions. In particular, by considering as a test bench a wheeled, non-holonomic COTS robot whose only accessible sensor is a low-resolution camera, previously unavailable localization and autonomous navigation capabilities are developed for it by exploiting dead reckoning and artificial landmark detection algorithms. These capabilities could then be exploited to create new types of games, which can be played in free-scale unknown environments and feature new forms of interaction. Implementation details of a sorting game in which the robot acts as a referee are reported.
- 10:30 Feature Based Analysis Friendly Video Coding Scheme for HEVC
- This paper proposes an algorithm to maintain feature information of compressed video in high efficiency video coding (HEVC), where feature information is defined by SIFT (scale invariant feature transform) detector. Adaptive largest coding unit (LCU) selection is defined to determine LCU into two different groups, important LCU group and non-important LCU group. Moreover, two different bit allocations are generated in rate control to each group based on coding mode selection, Intra or Inter mode, to achieve target bit rate and also guarantee the feature information can be maintained. Experimental results show proposed method can achieve better SIFT similarity than original HEVC encoder from 1% to 10% at the same target bit.
- 10:45 Bit Allocation and Encoding Parameter Selection for Rate-Controlled Error Resilient HEVC Video Encoding
- Even though the latest video compression techniques such as High Efficiency Video coding (HEVC) have succeeded in significantly alleviating the bandwidth consumption during high resolution video transmission, they have become severely susceptible to transmission errors. Overcoming the resulting temporal impact of the transmission errors on the decoded video requires efficient error resilient schemes that can introduce robustness features to the coded video in order to mitigate the negative impact on the viewer. To this end, this paper proposes a rate-controlled error resilient bit allocation scheme, together with an encoding parameter selection process, to adaptively determine the most robust video coding parameters and the decoder error concealment operations during the encoding itself. Consequently, the proposed method has demonstrated 0.48dB-0.62dB PSNR gain over the state-of-the art methods at the same bit rate.
- 11:00 Effective Natural Communication Between Human Hand and Mobile Robot Using Raspberry-pi
- We focus human-robot interaction (HRI) by performing natural communication between human hand and a mobile robot. The Raspberry-pi and Raspberry-pi camera module is installed on the robot. The system is capable to detect hand movement by processing the images from camera and respond according to the hand movements effectively. In addition, we set-up on-board LEDs react according each hand-movement to smoothen mutual communication between the robot and human hand. Through many experiments, we could confirm the effectiveness of the system.
- 11:15 Camera Orientation Estimation Using Motion Based Vanishing Point Detection for Automatic Driving Assistance System
- This paper presents a camera orientation estimation method based on 3-line RANSAC using motion vector in a driving straight ahead vehicle. The proposed method consists of three steps: i) motion vector and z-axis vanishing point estimation using feature extraction and matching, ii) line detection and classification using 3-line RANSAC algorithm, and iii) camera orientation estimation with vanishing points using classified lines. The experimental result shows proposed method effectively estimate camera orientation parameter. Therefore, the proposed method can be applied in vehicle systems for automatic driving assistance system.
- 11:30 An HEVC Fractional Interpolation Hardware Using Memory Based Constant Multiplication
- Fractional interpolation is one of the most computationally intensive parts of High Efficiency Video Coding (HEVC) video encoder and decoder. In this paper, an HEVC fractional interpolation hardware using memory based constant multiplication is proposed. The proposed hardware uses memory based constant multiplication technique for implementing multiplication with constant coefficients. The proposed memory based constant multiplication hardware stores pre-computed products of an input pixel with multiple constant coefficients in memory. Several optimizations are proposed to reduce memory size. The proposed HEVC fractional interpolation hardware, in the worst case, can process 35 quad full HD (3840x2160) video frames per second. It has up to 31% less energy consumption than original HEVC fractional interpolation hardware.
- 11:45 Decoding Complexity-Aware, Rate, Distortion Optimized HEVC Video Encoding
- Video content adaptation has become a popular application layer approach to reduce the decoding complexity and the associated energy consumption of video playback. In this context, this paper proposes an encoding algorithm to generate less complex HEVC bit streams with minimal impact to the coding efficiency. The experimental results reveal a -17.27 \% average decoding complexity reduction and up to -24.91 \% energy reduction for openHEVC based software decoding with only -0.73 BD-PSNR loss.
- 10:30 Optimizing Memory Swapping Scheme on the Memory Debugging Platform of CE Devices
- In this paper, we revisit the traditional memory debugging approach for CE Devices, and propose a novel memory debugging platform by modifying both DUMA library and Zram. Especially, we focus on the memory space wasted while detecting the memory error, because most CE devices have limited memory resource. For evaluation, we implemented a prototype of our debugging platform on Linux kernel version 4.8.17 and conducted various experiments with two different workloads. Our experimental results show that our scheme improves the performance of swapping by up to 58 times.
- 10:48 The Minimal-Effort Write I/O Scheduler for Flash-based Storage Devices
- Unfortunately, current I/O stacks in the operating system cause the read-blocked-by-write problem because they try to issue write requests that came from the applications in a burst way. In this paper, we propose a novel I/O scheduler to efficiently mitigate the problem. The key idea of our scheduler is the minimal-effort write that calibrates the number of dispatched requests from the queues of the I/O scheduler. The evaluation results clearly show that our scheme improves overall read latency while guaranteeing the throughput required by the application. In the best case, our scheme reduces average read latency by up to 66% compared to conventional I/O schedulers.
- 11:06 User-based Resource Scheduling for Multi-user Surface Computing Systems
- With technical advances in surface computing, surface computing devices such as tabletops support multi-user environments where multiple users run individual applications at the same time. To provide user-based scheduling and performance isolation among the users in such systems, we propose a new resource scheduling scheme for multi-user surface computing environments. The proposed scheme creates an individual resource group (user group) for each user, and inserts all processes of the user into the user group at the time the processes are created. Since the user group can be isolated from other user groups and the system can allocate resources based on users, not processes, the proposed scheme provides user-based scheduling and performance isolation among the users. In our experiments, the proposed scheme shows 1.56 times and 1.66 times better performance for CPU and I/O benchmarks, respectively, compared to the conventional scheme.
- 11:24 RFID-Based Scheme for TV Receiver Control in Case of Theft
- Cargo theft is a major problem in consumer electronics logistics chains, due to additional costs regarding security schemes, which try to avoid attempted robbery or track devices, and financial losses, which happen when criminals succeed in stealing goods. Although typical approaches, such as armed escort and electronic decoys, are somehow effective, they normally need additional support from other departments and do not make devices inoperable. The present article address such a problem and proposes an RFID-based methodology for disabling and enabling TV receivers, during transportation and when final destinations are reached, respectively. Real experiments revealed enhancement opportunities, allowed fine tuning, and finally showed its effectiveness, when individual devices are received, unloaded, and enabled, at costumer sites.
- 11:42 Design of Spectrum Shaping Codes for High-Density Data Storage
- We propose a systematic code design method for constructing efficient spectrum shaping k constrained codes for high-density data storage systems. Simulation results demonstrate that the designed codes can achieve significant spectrum shaping effect with only around 1% code rate loss and reasonable computational complexity.
(1) Melissa Sassi, Microsoft
(2) Maria Palombini, IEEE Standards Association
(3) Rudi Schubert, IEEE Standards Association
- 10:30 IEEE-SA Industry Connections -- Leading the Intersection of Emerging Technology and Policy
- Artificial Intelligence and Autonomous Systems are technology issues gaining substantial interest beyond the technical community, and into the minds of the general public. While IEEE has deep roots in the community of technical experts addressing these technologies, the IEEE Standards Association (IEEE-SA) has expanded the dialogue to include technology policy and ethical considerations that are increasingly impacted by these systems. The IEEE-SA Industry Connections program has been rapidly expanding to provide a platform for consensus building across a range of emerging technologies and their societal implications, incubating new opportunities for the IEEE to lead standards based solutions for new technology implementation. This panel session will address several of these new initiatives that impact the consumer electronics industry, as well as a range of additional industry applications. The IEEE Global Initiative for Ethical Considerations in Artificial Intelligence and Autonomous Systems has recently completed an updated version of its Ethically Aligned Design report. The first release of this report received widespread global attention and has been cited by the United Nations, World Economic Forum and many other major institutions for its thought leadership in a changing technological world. The IEEE Digital Inclusion through Trust and Agency initiative addresses agency over our data and cyber identity, along with internet related security considerations with a particular focus on blockchain applications and issues. This session will also introduce a new initiative on blockchain application considerations in the pharmaceutical and agriculture industries. Finally, via the IEEE Internet Initiative, activity is underway on standards considerations to support digital literacy and broader global inclusion in the internet and the digital economy. Standards activities have become a growing presence in developing structure and process around the societal implications of these topics, and IEEE has emerged as a leader with a dozen related standards projects launched in the past year, including the IEEE P7000 series, and more in development. The IEEE, through its global recognition as a community of experts, is helping to shape technology policy on these critical topics through standards and other activities to assure alignment with our mission of "Advancing Technology for Humanity".
Sunday, January 14 12:00 - 13:30
Lunch / Keynote 6: Proactive Cyber Security Response by Utilizing Passive Monitoring Technologies,
National Information Communication Technologies (NICT), Japan,
Yokohama National University
- 12:00 Proactive Cyber Security Response by Utilizing Passive Monitoring Technologies
- Recently, variety of cyber-attacks such as DDoS, Information Leakage, Illegal Access, Spam, Business E-mail Compromise, Phishing, Advanced Persistent Threats (APT), Man-in-the-Middle attacks are frequently recognized even in the consumer's environment. These cyber-attacks are often triggered by "malwares" and have been maliciously evolving and sometimes hidden from our monitoring countermeasures (FW, IDS/IPS). For proactively responding cyber-attacks, utilizing passive monitoring technologies should be reconsidered as possible security supportive solutions. In this talk, after introduction of latest cyber-attacks to share the current cyber threats landscape, passive monitoring technologies such as darknet and honeypot/sandbox will be explained with practical use-cases to accurately observe and monitor ongoing threats (cyber-attacks). The use-cases may include detection of malware-infected IoT devices by means of darknet and honeypot monitoring. Furthermore, detection of cyber-attacks by passive monitoring can be utilized for cyber security proactive response as practical solutions. Finally, future security considerations will be given for utilizing extendible passive monitoring technologies to proactively respond against cyber-attacks under smarter city and connected environments.
Sunday, January 14 13:30 - 15:00
- 13:30 Low-cost Authentication Paradigm for Consumer Electronics Within the Internet of Wearable Fitness Tracking Applications
- Nowadays, the consumer electronic (CE) market has grown to include smart devices for the Internet of Things. With increased adoption of these smart IoT devices, there is a high demand for safe and secure authentication for the purpose of maintaining effective access control. This paper addresses the concerns of authenticating consumer electronic devices using low-cost methods for implementation within customer healthcare frameworks. Our aim is not only to tackle the security issues related to the malicious behavior of supply chain counterfeit devices but also for establishing a secure architecture platform base for users to authenticate within a consumer electronic network. The proposed low-cost solution implements Physical Unclonable Function (PUF) for device verification and incorporates human behavioral characteristics (Biometrics) for user authorization as a comprehensive codesign method.
- 13:48 Improved Durability of Soil Humidity Sensor for Agricultural IoT Environments
- Soil humidity is the most important factor for plant growth. Therefore, the soil humidity sensor is an important part of smart farm application using agricultural IoT environments. Since soil humidity sensors are applied wet underground and the sensor consists of copper, rust eats away the copper surface of sensors. From rusting of sensors, wrong information of soil humidity can be collected on smart farm system based on agricultural IoT Environments. It makes that smart farm is not reliable. In this paper, we propose a new type of soil humidity sensor in order to extend life time.
- 14:06 On Analysis of Substitutability for System Resilience in IoT System Based on PN2
- We focus on a property to deal with system resilience in the Internet of Things system called substitutability. It is important to ensure that production line failure is avoided. Substitutability is a property of a machine or device to operate instead of another machine or device. In this paper, we modeled IoT by PN2 (Petri nets in a Petri net). We formalized substitutability problem and proposed a formal analysis method for the problem. We illustrate the analysis with a substitutable and non-substitutable example of a drone worker in a smart warehouse. The effectiveness is shown as the accuracy to equate the difference of behavior in agent substitutions and to reduce the range of state space analysis of the whole PN2 to only the agent nets.
- 14:24 On the Cybersecurity of m-Health IoT Systems with LED Bitslice Implementation
- The Internet of Things (IoT) will provide a large scale infrastructure that can support a plethora of new networked services. One of the critical services is m-Health IoT which allows monitoring the health status of patients while providing the ability for a rapid response in emergency cases. Connecting healthcare services to the Internet through IoT brings new security threats and vulnerabilities that can jeopardize the patients' private data. In this paper, a novel security framework for m-Health IoT security is proposed using the concept of moving target defense (MTD). MTD allows the m-Health system to dynamically change its keys to increase uncertainty on an attacker and secure the data. In the proposed scheme, the devices update their keys locally to eliminate the risk of revealing the new keys while being shared with the gateway. A practical implementation is proposed based on bitslicing LED, a lightweight encryption cipher, to improve the performance of decrypting multiple packets at the same time. LED bitsliced implementation was tested on ARM Cortex-A53 which was shown to consume half of the processor's instructions compared to the original implementation. The effect of applying MTD on the number of processor's instructions is evaluated and shown to be bounded.
- 14:42 An IoT-Ready Streaming Manager Device for Classroom Environments in a Smart Campus
- The Internet of Things is slowly becoming reality. The proliferation of smart devices will take the human-machine relationship to a new level, and the classroom is not the exception. This paper presents the design and implementation of an IoT-ready classroom device for screen streaming based on an inversed VNC approach over TCP/IP on a wireless local area network. Streaming permission is granted following an authentication-authorization scheme to avoid undesired content streaming, thus allowing the operation over a public network. The device communicates to the network by publishing its status to a MQTT broker, which makes it an active thing in an IoT setup.
- 13:30 Dynamic State-Predictive Control for a Remote Control System with Large Delay Fluctuation
- Under the concept of IoT, various machines are beginning to be controlled remotely through the Internet and wireless IP networks. However, such an IP network brings about large delay fluctuation due to the nature of its packet queue. The delay is called dead time in control theory and makes it difficult to stabilize remote control. In this paper, we propose a novel state-predictive control that compensates for dead time by adjusting the prediction horizon dynamically on the basis of a real-time estimation of the time-varying delay. A simulation using actual delay data shows that the proposed method improves the stability and robustness.
- 13:52 Knowledge of Things: A Novel Approach to Share Self-Taught Knowledge Between IoT Devices
- The heterogeneity of Internet of Things (IoT) has increased as the number of IoT devices connected to the Internet has grown. This trend increases complexity and inconvenience for the use of various IoT devices. In this paper, we propose a Knowledge of Things (KoT) framework which facilitates sharing of self-taught knowledge between heterogeneous IoT devices which require similar or identical knowledge. The proposed model allows IoT devices to efficiently produce, cumulate, and share their self-taught knowledge with other devices in the vicinity.
- 14:15 System Architecture for IoT Services with Broadcast Content
- The development of smartphones and IoT devices may increase the touch points of broadcast content. Herein, we consider providing broadcast contents in a new way using a diversity of devices. We propose a system architecture and an application framework based on Hybridcast to communicate the metadata of broadcast content among a TV, a smartphone, and IoT devices. Moreover, we prototype two use cases and indicate that two kind of services are realized with one framework based on our proposed system.
- 14:37 Versatile Prediction Core for IoT Applications
- A novel prediction method for IoT applications is presented. The proposed method estimates both the state and the state change of the tracked targets by processing multidimensional features from numerous external sensors using machine learning. The proposed method achieves longer-term prediction with less computational time than conventional methods.
- 13:30 Robust Photometric Alignment for Asymmetric Camera System
- Multiple image sensors have now become the defacto offering from all major mobile phone manufacturers. Their applications vary from improving the image quality to providing new features to the end-users. One of these multi-sensor system is a tele-wide asymmetric camera configuration which is aimed at providing better zoom experience with more crisper images. One such major challenge is accurate photometric alignment which is required to compensate the difference in auto exposure(AE), auto white balance(AWB), and the inherent differences in terms of both intrinsic and extrinsic parameters of the two camera sub-systems. Conventional methods in literature for photometric alignment make several assumptions and are also not very adaptable to different lighting conditions. To the best of our knowledge, accurate photometric alignment intended towards fusion of images from asymmetric tele-wide cameras that works in unconstrained scenarios and at all zoom ratios, is not very well studied. In this paper, we propose a computationally efficient algorithm for brightness and color correction that is suitable for fusion of images from multiple asymmetric cameras. The proposed method is evaluated quantitatively using the RMSE metric and qualitatively.
- 13:45 Multiple Feature Instance Detection with Density Based Clustering
- This paper describes an approach to local feature instance detection by treating this as a clustering problem. We use a density based clustering algorithm and a local feature detection algorithm. The paper also provides an approach to validating the results. Confidences are calculated for each cluster using core distances and intra-cluster distances to determine how valid the each cluster is for a particular value of the clustering algorithm parameter for minimum cluster size. We used the statistical values of kurtosis and skewness to validate the choice of minimum cluster size and describe a way of using these values to make an informed choice of the proper value of minimum cluster size.
- 14:00 Adaptive Noise Canceller with Two SNR Estimates for Stepsize Control
- This paper proposes an adaptive noise canceller with two SNR estimates for stepsize control. The first SNR, which is a power ratio of the error signal and the adaptive filter output, is used conventionally for stepsize control. The second SNR is newly introduced as a power ratio of the primary and the reference input signals. The coefficient adaptation stepsize is basically controlled by the first SNR. However, in the initial period when a high first SNR prohibits adaptation, the second SNR is used as an approximation to the first SNR. Switching from the second to the first SNR is performed when the latter takes a smaller value than the former. Evaluations with a speech mixture demonstrates that use of the second SNR followed by the first SNR successfully made the sum of squared coefficients approach the true value which represents convergence.
- 14:15 An Efficient FPGA Implementation of HEVC Intra Prediction
- Intra prediction algorithm used in High Efficiency Video Coding (HEVC) standard has very high computational complexity. In this paper, an efficient FPGA implementation of HEVC intra prediction is proposed for 4x4, 8x8, 16x16 and 32x32 angular prediction modes. In the proposed FPGA implementation, multiplications with constants are implemented using DSP blocks in FPGA. The proposed FPGA implementation, in the worst case, can process 55 Full HD (1920x1080) video frames per second. It has up to 34.68% less energy consumption than the original FPGA implementation of HEVC intra prediction. Therefore, it can be used in portable consumer electronics products that require a real-time HEVC encoder.
- 14:30 Optimization of HEVC Lambda-domain Rate Control Algorithm for HDR Video
- Rate Control (RC) will play an important role in conceiving high-fidelity HDR video distribution through transmission and broadcast in imminent future. However, as video coding standards, like HEVC, are generally designed and optimized for the LDR content, this can result in an inefficient HDR video compression in the Rate- Distortion (RD) sense. In this paper, the state-of-the-art HEVC Lambda-domain RC algorithm is optimized for HDR video coding by proposing a new Lambda-QP relation after investigating the suitable RD model for HDR content. The updated RC algorithm with proposed relation outperforms the default HEVC RC algorithm achieving averaged PU-PSNR improvements of up to 1.36 dB for HDR video test sequences. Further, performance improvement is also evident from the HDR-VDP-2.2 Q and HDR-VQM based RD performance curves.
- 14:45 HDR Image from Single LDR Image After Removing Highlight
- A low dynamic range (LDR) image that contains highlight does not provide proper information at the highlight area. Besides, all area of the image may not be exposed properly. Removing the highlight and then converting the image to the high dynamic range (HDR) image will increase the quality of the image from visual perception. In this paper, we propose a method for removing the highlight from an LDR image and convert the highlight free image to HDR image by using tone mapping operator (TMO). After detecting highlight area of an image, modified specular free (MSF) image is used to remove highlight part from the LDR image. Then, the highlight free LDR image is converted to HDR image by TMO. Finally, we have measured the quality of our output image to show that output image has better dynamic range than the input image.
- 13:30 Robust Bounded-Error Subset Selection
- We show that the bounded-error subset selection problem can be interpreted as a worst-case robust optimization problem, where the solution vector is robust under worst-case signal perturbations. In particular, the solution to the subset selection problem is robust when the signal under consideration is allowed to take arbitrary values within a fixed-interval uncertainty set. We also show that the robust behavior of the bounded-error subset selection is superior than the well-known methods for solving the subset selection problem (e.g., Basis Pursuit, Orthogonal Matching Pursuit, and the Lasso algorithms)
- 13:45 Transformer-DARwIn: A Miniature Humanoid for Potential Robot Companion
- Robot companionship has become more popular in past years. However, humanoid gait might be somewhat unstable for these applications. Even with miniature humanoids, falls occur frequently. Thus, wheel attachments have been added onto a miniature humanoid, so it can move faster and with more stability than walking. In addition, with such attachments a robot can switch from walking to rolling when necessary. DARwIn-OP is a humanoid robot that has been used as an experimentation and performance evaluation platform. This paper discusses preliminary work regarding robot companionship applications by using a miniature humanoid capable of fetching different toys based on voice command.
- 14:00 Direct Canvas: Optimized WebGL Rendering Model
- WebGL has been introduced as one of the powerful web features for developing 3D content. However, there are still performance issues when compared to the native GL applications. The traditional WebGL rendering requires unnecessary buffers and copies even though when WebGL is the only content in the current web page. This paper presents an optimized rendering model when WebGL is used as dominant content like 3D games. And we will demonstrate its effectiveness with performance data measured on the embedded devices.
- 14:15 Mitigating Write Interference on SSD in Home Cloud Server
- Recently, some home appliances such as smart TVs and home gateways adopt virtualization technologies to perform the role of home cloud server and govern all connected devices efficiently. In such virtualized systems, it is important to satisfy the I/O performance SLO (Service Level Objective) of each VM. In this paper, we propose a novel write buffer management scheme for SSDs in home cloud server, which guarantees the SLO of each VM by mitigating write interference problem among the VMs. Experimental results clearly show that our scheme outperforms the conventional write buffer scheme in terms of I/O performance SLO of each VM.
- 14:30 Smart Coffee Roaster Design with Connected Devices
- The paper proposed a smart coffee roaster design with connected devices. The proposed design provides the bean temperature, bean cracks and carbon monoxide monitoring and also allows the operator full control even at a distance from the unit and isolates the operator from potentially harmful side effects such as headache, dizziness, etc. Moreover, the proposed design not only makes the device safer for the patron to use but also paves the way for automating other home appliances like an oven, washer and dryer, etc.
- 14:45 Mass Production EMC Status Quick Evaluation Through Cloud Computing - A New Option for Electric Product Quality Control
- This paper proposes an innovative methodology for mass production EMC status quick evaluation. The evaluation is carried out in workshop instead of a shielding room, with a digital oscilloscope to acquire product data which is further calculated by Cloud Computing. An application example of the proposed methodology is introduced which not only indicates that product with fail EMC status may be discovered from mass production quickly in low cost, but also unveils there is certain relationship between product quality and EMC status.
Panel Chair: <a
Senior Program Director, IEEE Future Directions
(1) <a href="https://cesoc.ieee.org/images/files/images/nicholas.jpg">Nicholas Napp, Xmark Labs, LLC
(2) Conor Russomanno, Metavision
(3) Joe Sullivan, The Weather Company
(4) Tao Zhang, Cisco Systems
(5) Tom Coughlin, Coughlin Associates
- 13:30 "Immersive Future" with New and Emerging Innovations and Predictions
- The session hopes to demonstrate the interplay and intersections among Big Data, Brain, Digital Senses, and Fog Computing. Emphasis will be upon presenting practical applications and its implementations. Invited subject matter expert speakers will comment on current and past implementations, with a focus on predictions of the future and potential implications to the consumer. In particular on aspects of AR, VR, and Mixed Reality.
Sunday, January 14 15:00 - 15:30
Sunday, January 14 15:30 - 16:00
- 15:30 Adaptive Remote Control of a Mobile Robot System with Delay Fluctuation
- Since the concept of an Internet of Things (IoT) has become widespread, remote control of IoT devices by long-distance communication through the Internet and wireless networks is gaining much attention. However, best-effort IP networks such as the Internet and wireless IP networks bring about a time delay fluctuation because of cross-traffic and radio interference. The delay fluctuation deteriorates the stability of remotely controlling IoT devices such as mobile robots, which is a major challenge of a long-distance remote control system. In this paper, we construct an experimental environment for long-distance communications using the Internet and a wireless LAN and investigate round trip times (RTTs) on it. Furthermore, we present a novel remote control method for a mobile robot by dynamically adjusting its moving velocity on the basis of a real-time estimation of a network delay. A simulation of the remote control using the obtained RTT data shows that the proposed remote control achieves highly efficient movement of the robot even on a network that has time delay fluctuation.
- 15:31 Self-Configuration Without Commissioning for Wireless Internet of Things Using Subjective Logic
- Internet of Things (IoT) applications need commissioning and configuring of the IoT infrastructure that is comprised of devices called 'Things'. In some cases, e.g. for an intelligent street lighting infrastructure in a city, we are looking at tens of thousands of Things that need to be commissioned and configured. This procedure can take several months, is very costly and error prone. This becomes especially tricky with increased use of wireless communication thanks to deployment that does not require wiring. For IoT applications that employ reasoning based on location, we propose a low-complexity and adaptive method based on computational trust for self-(re)configuration of wireless Things. The proposed method does not require commissioning.
- 15:33 Measurement and Evaluation of Comfort Levels of Apartments Using IoT Sensors
- We developed IoT sensors for the evaluation of the comfort level of real estate building properties. An IoT sensor can measure temperature, humidity, illuminance, and acceleration and can record audio and video. We measured these elements in rooms using the IoT sensors. From this experiment, we succeeded in evaluating the difference in comfort levels of the rooms. For example, we determined quantitatively that an upper-floor room received more sunshine than a lower-floor room despite the fact that the two rooms were in the same building.
- 15:34 A Study on Development of the Blind Spot Detection System for the IoT-Based Smart Connected Car
- The Internet of things (IoT) is an emerging topic in a variety of industry fields such as manufacturing, engineering, automobile, etc. Especially, many automobile companies are competitively introducing a smart-connected car in conjunction with the IoT technology. For instance, some recent vehicles support autonomous driving using cameras and radar-based safety features such as pre-collision prevention system, lane departure alert with a steering assistant function, etc. Based on this example of the IoT-based smart connected car, we actually developed the camera-based blind spot assistance system. The power / serializer board and camera board consisting of a number of sensors, processors, and other hardware components were developed by our own system interface design, and we confirmed that the developed system successfully met quantitative evaluation criteria through testing. Furthermore, we are proposing an idea of object detection in the blind spot based on a deep learning methodology, as well as proposing an embedded system design for transmitting the acquired image data to the server to fulfill the connectivity characteristic.
- 15:36 Census: Continuous Posture Sensing Chair for Office Workers
- Continuously undesirable or same sitting position for a long time causes the burden on the waist and shoulders, leads to mental stress, reduces work efficiency. Therefore, it is important to measure their sitting situation, improve the sitting pose, change the posture periodically, and move from the seat. At that time, it is necessary to continuously measure without affecting the work of the office worker. Consequently, we developed a sensing chair called ''Census'' satisfying those conditions. This sensing chair has a function to classify the sitting posture of workers with the accuracy of 80.2%.
- 15:38 Lightweight Smart Home Security System Using Multiple RSS-based Voting
- A centralized smart home system can be attacked by hacking the center or camouflaging as a member device. We propose a distributed security system using received signal strength (RSS) based voting. The peripheral devices vote according to whether the RSS is in a pre-calculated range. The validation of a data is conducted by the result of vote. If the peripheral devices approve more than 60%, the data is regarded to be trusted. We have confirmed through actual implementation and experiments against the camouflage attack scenario. As a result, the proposed system blocked the data from a camouflaged device as 100% in the experiments.
- 15:39 A Smart Middleware to Detect On-Off Trust Attacks in the Internet of Things
- Security is a key concern in Internet of Things (IoT) designs. In a heterogeneous and complex environment, service providers and service requesters must trust each other. On-off attack is a sophisticated trust threat in which a malicious device can perform good and bad services randomly to avoid being rated as a low trust node. Some countermeasures demands prior level of trust knowing and time to classify a node behavior. In this paper, we introduce a Smart Middleware that automatically assesses the IoT resources trust, evaluating service providers attributes to protect against On-off attacks.
- 15:41 A Security Evaluation of Popular Internet of Things Protocols for Manufacturers
- The Internet of Things (IoT) continues to increase in popularity as more "smart" devices are released and sold every year. Three protocols in particular, Zigbee, Z-wave, and Bluetooth Low Energy (BLE) are used for network communication on a significant number of IoT devices. However, devices utilizing each of these three protocols have been compromised due to either implementation failures by the manufacturer or security shortcomings in the protocol itself. This paper identifies the security features and shortcomings of each protocol citing employed attacks for reference. Additionally, it serves to provide recommendations for secure implementation of these protocols.
- 15:42 Development of Laundry-work Assistance Robot by Using IoT Technology
- In this paper, we developed a laundry uptake robot that supports tasks such as "dry" and "take in" for the laundry work which is one of "Instrumental Activities of Daily Living". This robot has mainly three functions, which protect laundry from sudden rain by using rainfall prediction information and water sensor, automatically take in laundry, and control these operations remotely. Battery charged from a solar panel was used to reduce power consumption. With this robot, we can dry laundry outside without worrying about rain, and we can take in the laundry wherever we are.
- 15:44 Non-Invasive Real Time Continuous Monitoring of Blood Glucose Smart Sensor Design Based on IoT
- Diabetes, a really serious and progressively common disease, happens once the body cannot able to use insulin properly to transfer glucose to the cells that required for energy. This condition leads to blood glucose levels being unreasonably high or hazardously low, these may cause to serious complications like visual disorder, kidney disease, amputation and vascular disease. Proper precaution should be taken at soonest possible. So, there's a strong requirement of non-invasive type Continuous glucose monitoring system. This paper presents a cost effective, low-power, non-invasive Continuous glucose monitoring system for analysis of blood sugar levels in the body. The system is based on Internet of Things and cloud computing. Internet of Things is the network of internet connected objects which are capable of collecting and exchanging data with the help of embedded sensors. The system is using low cost open source Arduino Uno Board which collects the data coming from sensors unit and transmits the data to the cloud using Wifi shield. The real-time data sent to the cloud can be monitored in the form of labeled data and graphs anytime on blynk app installed on Android or iOS based mobile phones having internet connectivity.
- 15:46 Five Acts of Consumer Behavior: A Potential Security and Privacy Threat to Internet of Things
- The existing security and privacy preserving solutions proposed for IoT-enabled consumer products overlook communal acts of a consumer behavior. They lack support in case when IoT Consumers borrow, rent, gift, resale and retire their IoT-enabled electronic products. In this paper, we attempt to highlight IoT consumer's security and privacy violation through seemingly five different angles. We also speculate that IoT products or devices, even after their End of Life (EOL), i.e. "IoT waste", could become an attractive gateway for cyber criminals to access private information. Consequently, we present some recommendations in this regard.
- 15:47 Study of Localization Method for Switching Between Low Electricity Consumption and High Precision for a Watching System
- We propose a hybrid localization system for watching security. In a usual monitored area, the system localizes provided with high accuracy with relatively high power consumption. However, if there are abnormalities such as someone being taken or wandering, it switches to localization using Bluetooth Low Energy that can operate at low power consumption over a long period of time. As a result of experiments with actual machines, our proposal system estimates a terminal is within 5m became 100%. Moreover, the position was able to be estimated at 84.2% by using the finger print method.
- 15:49 Economic Incentive Based Solution Against Distributed Denial of Service Attacks for IoT Customers
- The market of Internet of Things is increasing at a very high rate. As most of the devices get connected to the internet, the security challenges of these devices are also increasing. With the abundance in the internet connectivity, distributed denial of service attack incidents is many folded in the recent past. The constrained nature of IoT devices makes them the easy target for the attackers. Further, the trade-off between quantitative factors like size, space and response time with a qualitative property like security mechanism, on a manufacturer's point of view, always prefers the quantitative properties and the security mechanisms in most of the IoT-enabled smart devices are minimal. Considering such a scenario, we propose a model to secure the IoT network like a smart home using a secure gateway router. Further, we propose the risk transfer mechanism based on economic incentive based agreement. It can be the most feasible solution in a case if the owner of the network is not always able to ensure the security of the whole network. An already communicated third party will take care of the security in case of any security breach.
- 15:50 A Robust Anonymity Preserving Authentication Protocol for IoT Devices
- In spite of being a promising technology which will make our lives a lot easier we cannot be oblivious to the fact IoT is not safe from online threat and attacks. Thus, along with the growth of IoT we also need to work on it's aspects. Taking into account the limited resources that these devices have it is important that the security mechanisms should also be less complex and do not hinder the actual functionality of the device. In this paper, we propose an ECC based lightweight authentication for IoT devices which deploy RFID tags at the physical layer. ECC is a very efficient public key cryptography mechanism as it provides privacy and security with lesser computation overhead. We also present a security and performance analysis to verify the strength of our proposed approach.
- 15:52 Efficient Management of Precision Agriculture Using Wi Fi Based Sensor Nodes and IoT Services
- This paper proposes a system for the management of an agricultural environment using Wi-Fi based Wireless Sensor Networks (WSN) and Internet-of-Things (IoT). The system collects data from the environment and soil through the sensor nodes in the network. The collected data is stored on IoT servers and the farmer can monitor and control his farm such as turn on valves, motors etc, receive alerts and notifications updates from anywhere in the world through his farming website and also through social networking sites. What is unique in this paper is the way we have used WiFi and IoT services. We have used the most recent, efficient and low power hardware and software for the realization of our goals. We have also used the services offered by many kinds of IoT service providers to make this work attractive and useful. This is a big change from established agricultural procedures. But it is expected that such innovative technologies when introduced into the farming sector will attract the younger generation into this profession by making it more productive and easy to manage.
- 15:54 Dynamic Autonomous Frequency Reuse for Uplink Cellular Networks
- In this paper, a dynamic autonomous frequency reuse approach is adopted for resource allocation in the uplink of cellular networks. It is assumed that each cell is divided into inner and outer regions. The proposed approach is based on minimizing the total uplink interference at every eNB for all users in both inner and outer regions in the network. The main advantage of the proposed scheme is that it adapts itself to the varying wireless channel. Simulation results show that the proposed approach provides better total network uplink throughput and average user uplink bit rate compared to the traditional fractional frequency reuse (FFR) regardless of the User Equipment (UE) transmit power, cell radius and user density within the cell.
- 15:55 Study on Headphone Hearing Loss Prevention Methods Based on the Melody Structure of Music on Portable Music Player
- In this paper, we examine the volume control system in an application to prevent the onset of headphone hearing loss due to music listening using a portable music player. First, the possibility that lower the volume to a safe level without be noticed by the user (difference of perceived attenuating level) are evaluated. Based on the above experiments, we examine the melody structure of music by attenuate the volume separately for melody structure (Intro part, Verse, Chorus, Interlude). Finally, we evaluated difference of the perceived attenuating level considering frequency characteristics.
- 15:57 Single-Channel Speech Dereverberation Based on Block-wise Weighted Prediction Error and Nonnegative Matrix Factorization
- This paper proposes a speech dereverberation method based on a block-wise weighted prediction error (BWPE) method and nonnegative matrix factorization (NMF). The proposed BWPE method iteratively estimates late reverberation using maximum likelihood (ML) estimation in a block-wise manner. To ensure consistent dereverberation performance over time, a forgetting factor is applied on intermediate estimates. Thus, the recent statistics of the signal are emphasized during the block-wise processing. In addition, the NMF-based source separation meth-od is applied to reduce early reverberation that remains in the signal processed by the proposed BWPE method. The perfor-mance of the proposed method is compared with that of the con-ventional weighted prediction error (WPE) method by measur-ing the Segmental signal-to-noise ratio (SSNR). It is shown from the comparison that the proposed method achieves a higher SSNR than the conventional method. Moreover, the proposed method can be implemented in a real-time audio recording de-vice with an algorithmic delay of 20ms.
- 15:58 Portable Intelligent Home Service System Based on SSVEP
- In this paper, a portable intelligent home service system based on Steady State Visually Evoked Potential(SSVEP) is constructed. This system is directly controlled by the brain, which is very suitable for special environment operations and assistive activities for the disabled. It is combined with intelligent home to meet the needs of the disabled and improve the quality of life of the disabled. Compared with the traditional BCI system, this system has better portability and cost performance. We also analyze the performance comparison between the Canonical Correlation Analysis(CCA) algorithm and the classical Fast-Fourier transform(FFT) algorithm under different sampling time in this paper. The experiment results demonstrate that the system has the characteristics of portable, small volume, low cost and high accuracy.
Sunday, January 14 16:00 - 17:00
- 16:00 Simple Online and Realtime Tracking with Spherical Panoramic Camera
- In this paper, a simple yet effective method for online and realtime multi-object tracking (MOT) on 360-degree panoramic videos is proposed. Based on current state-of-the-art tracking-by-detection paradigm, several improvements have been made to overcome the challenges of full field-of-view (FOV) of Spherical Panoramic Camera (SPC). In addition, two datasets are presented for evaluation. It is shown that the proposed method outperforms the baseline by 28.6% and 27.8% in terms of average Multiple Object Tracking Accuracy (MOTA) on each dataset.
- 16:20 Development of Projection LED Guiding Sign for Dynamic Motion Imaging
- Recently, the flow lines or the lines of action have been getting knotty in railway stations, public establishments and office buildings as the facilities and service outlets operating there are complexly organized and the security control is more enhanced. Although a variety of signs are provided there to guide people and goods, they are considered insufficient presently because it is sometimes difficult to reach intended destinations smoothly. In order to provide a solution to the issue, this paper proposes a projection LED guiding sign that projects guiding signs on the walls and floors.
- 16:40 Similar Floor Plan Retrieval Featuring Multi-Task Learning of Layout Type Classification and Room Presence Prediction
- In this paper, a new framework for real estate property search using a floor plan image as a query is presented. In similar property search, appearance-based similar floor plan image retrieval does not work well because similar properties would have totally different floor plan images. Therefore, we propose a multi-task learning using deep neural networks to solve this problem. In our proposed method, Convolutional Neural Networks (CNNs) are trained to solve the two tasks: one is layout type classification and the other is room presence classification.
- 16:00 DNSLedger: Decentralized and Distributed Name Resolution for Ubiquitous IoT
- With the popularity of internet of things, object identifier has gradually been taken seriously. As the identification information of the object, object identifier has many similarities with internet domain name. Despite the current domain name system is already mature enough, there are still some fundamental problems. Aiming at altering the centralized domain name management model, we propose a decentralized domain name system-DNSLedger. DNSLedger is a distributed system based on blockchain, which implements the features of common maintenance, node equivalence and so on. This paper will make a brief introduction to DNSLedger and analyze what advantages the proposed model has. To provide a feasibility scheme for the management of object identifier.
- 16:20 Real-time Speaker Recognition System Using Multi-stream I-Vectors for AI Assistant
- We describe a real-time speaker recognition that works in sync with ASR engine. In order to improve recognition accuracy, we employed multi-stream i-vectors. The proposed multi-stream i-vectors compensate the limited amount of information contained in a single i-vector. Our approach provides superior performance compared with GMM based approach.
- 16:40 Ensemble Based Real-time Indoor Localization Using Stray WiFi Signal
- Accurate Indoor localization using minimalist hardware has a huge potential for energy savings and smart homes. Current indoor localization techniques involve deployment of complex array of sensors such as Bluetooth beacons, PIR sensors, etc. However, these sensors are commonly intrusive and require extra hardware which might not be available in most homes especially in underdeveloped countries. In this paper, we present an ensemble based learning technique to detect a real-time room level location inside a home situated in a residential complex. The method only uses existing WiFi infrastructure in the complex. The method uses an ensemble of classifiers on a weighted time averaged RSS signals to detect the user location. The analysis was done on the data collected by authors in a residential setup through a self-developed android application. The home setup included 4 locations (2 Bedrooms, Living Room and Kitchen). The total prediction accuracy achieved was 85.7%. The localization algorithm was reaffirmed on the Dutch Residential Energy dataset (DRED) and is also compared to those obtained using multiple bluetooth beacons used for localization using the DRED dataset. These localization results can be further utilized to impart better intelligence to smart home devices.
- 16:00 User-friendly Inter-Pupillary Distance Calibration Method Using a Single Camera for Autostereoscopic 3D Displays
- The accurate inter-pupillary distance (IPD) of a user plays an important role and is a prerequisite for eye-tracking-based autostereoscopic three-dimensional (3D) display systems by calculating the precise 3D eye position of the users. We aimed to develop a robust computer-aided algorithm for each user-specific IPD calibration using a single camera in a user-friendly manner. Our algorithm consists of eye tracking, pattern rendering, user pattern selection, and IPD adjustment according to the selected patterns. Two stereo patterns were designed to clearly show the IPD differences: a 3D stereo registration pattern and a complimentary stereo pattern. We applied this algorithm to 21 users. The reference standard was provided by a commercial pupilometer. The IPD values obtained by the proposed method and the reference standard IPD values were not statistically different (64.9 ± 4.1 mm from the algorithm and 64.2 ± 3.4 mm from the reference standard, p = 6.64) from the students' t-test. A good agreement was observed among the 21 users in using the IPD calibration software with an agreement of 94.8% (kappa 0.89, 95% confidence interval from 0.83 to 0.96, and p < 0.0001). Our algorithm shows promising results in IPD calibration using a single camera in a user-friendly manner.
- 16:15 Robust, Precise, and Calibration-Free Shape Acquisition with an Off-the-Shelf Camera and Projector
- Recently, three-dimensional (3D) shape acquisition systems composed of simple commercial devices have received significant attention from both professional and nonprofessional users. In this work, we propose a flexible projector-camera system that can accurately acquire whole shapes of 3D objects without requiring any special calibrations. The proposed system is based on structured-light and Structure-from-Motion (SfM) techniques that use coded patterns as dense discriminative features instead of the local features popularly used in SfM. To this end, recordings of coded patterns are inserted by moving the camera (to capture image data from multiple camera view points) and projection of the patterns by moving the projector. The proposed system can accurately recover even texture-less objects owing to the advantages of both SfM and structured-light. We demonstrate the benefits of this system by several experiments using real data and compare it to current methods.
- 16:30 Mosaic Block Detection Based on HOG with SVM Classifier and Template Matching
- Digital video signal is widely used in modern society. There is increasing demand for it to be more secure and highly reliable. Focusing on this, we propose a method of detecting mosaic blocks. Our proposed method combines two algorithms: HOG with SVM classifier and template matching. We also consider characteristics of mosaic blocks other than shape. Experimental results show that our proposed method has high detection performance of mosaic blocks.
- 16:45 An HEVC Real-time Encoding System with High Quality HDR Color Representations
- In recent years 4K video broadcasting and distribution services have spread rapidly and high dynamic range (HDR) technology has been attracting a great deal of attention to make its 4K video services more realistic. We have developed and proposed new coding control methods to improve perceptual quality by reducing degradation of HDR-specific color artifacts. The proposed methods improve the video quality degradation by color difference on average 2.35% in the most degraded frames, while PSNR remains almost same.
- 16:00 EMI Designs Support System Using Augmented Reality
- In this paper, we introduce the development of Electromagnetic interference (EMI) designs support system using Augmented Reality (AR). First, we describe the development efficiency improvement effect and the advantages not depending on the Equipment Under Test (EUT) size by using AR technology for the visualization system of the near electromagnetic field noise. Then, we explain the wave source model inverse problem analysis approach, which estimates the radiation mechanism of the unnecessary radiation noise based on the measurement value of the near electromagnetic field noise.
- 16:20 Road-illuminance Level Inference Across Road Networks Based on Bayesian Analysis
- This paper proposes a road-illuminance level inference method based on the naive Bayesian analysis. We investigate quantities and types of road lights and landmarks with a large set of roads in real environments and reorganize them into two safety classes, safe or unsafe, with seven road attributes. Then we carry out data learning using three types of datasets according to different groups of the road attributes. Experimental results demonstrate that the proposed method successfully classifies a set of roads with seven attributes into safe ones and unsafe ones with the accuracy of more than 85%, which is superior to other machine-learning based methods and a manual-based method.
- 16:40 Indoor Localization Using Visible Light Communication and Image Processing
- Indoor localization using short-range wireless communication techniques has gained popularity over the Global Positioning System due to the latters limited capability to provide indoor positioning information. This paper reports the implementation of a multi-transmitter visible light communication based indoor localization system that offers a moderate data rate and indoor positioning with sub-meter accuracy. The work includes a prototype of the proposed system with four transmitters and a receiver module mounted on a stationary base. The transmitted data was Manchester encoded, and Binary Amplitude Shift Keying modulated. Multiple access of channel was achieved using Time Division Multiple Access technique. The receiver module used a PIN photodiode to detect the light signal and indoor positioning or localization was implemented using received signal strength technique.
(1) Tom Coughlin, IEEE CE Society, email@example.com
(2) Lee Stogner, IEEE CE Society, firstname.lastname@example.org
(3) Soumya Kanti Datta, Future Tech Lab, India, email@example.com
- 16:00 IEEE CE Society Future Directions Session
- This is a proposal for a session on IEEE CE Society Future Directions Session. The mode of presentations will be oral followed by a question and answer session. This is organized by the FD-IoT team.
Sunday, January 14 17:00 - 17:30
- 17:00 A Frequency-shift Readout System with Offset Cancellation OPA for Portable Devices of Marijuana Detection
- This paper presents a novel frequency-shift readout system with offset cancellation OPA (operational amplifier) for biosensing applications. The chopper technique is used to up-modulate the offset voltage to a higher frequency for noise shaping. A low pass filter is then utilized to filter out the noise caused by offset. A FPW (flexural plate wave) biosensor to sense the THC (Tetrahydrocannabinol), the signature ingredient of marijuana, concentration in urine. The linearity of the proposed system is proved to be 0.959 based on physical experiments where the THC concentration is 0 ~ 160 ng/mL.
- 17:01 A Prototype Foot-Sole Acetone Analyzer and Its Applicability to Caloric Balance Monitoring
- The authors of this paper prototype a foot-sole acetone analyzer world first which can measure skin emitted acetone by having the user simply step onto the prototype like a weight scale. Our results revealed that plus or minus of the total caloric balance can be estimated from the amount of emitted foot-sole acetone with 79% accuracy. This result indicates that there is no need to record caloric intake and caloric consumption, and foot-sole acetone measurements by using our prototype could be useful for dieting management at home.
- 17:03 A Safety Engineering on the Design of Hemodialysis Systems
- Consumer electronics systems are being largely used for healthcare. In this context, we are developing an intelligent system for management and monitoring of hemodialysis treatment. However, this kind of system is considered critical because exposes risks to patients and users' lives. For this system, we need to assure that the patients will not to be exposed to risks. Hence, it is necessary to investigate the hazards present in the hemodialysis system that could represent critical risks to the patient. In this work, we identify and classify the hazards of the hemodialysis system as an important step to increase patients' safety.
- 17:04 Real-Time Cow Action Recognition Based on Motion History Image Feature
- In this paper, a cow behavior recognition algorithm is proposed to detect the optimal time of insemination by using the support vector machine (SVM) classifier with motion history image feature information. In the proposed algorithm, area information indicating the amount of movements is extracted from MHI, instead of motion direction which has been widely used for person action recognition. In the experimental results, it is confirmed that the proposed method detects the cow mounting behavior with the detection rate of 74%.
- 17:06 I-Jack: Wearable System for Collection and Evaluation Physiological Data
- This work presents a wearable system for collecting, storing and evaluating the physiological data of physical exercise practitioners. The system was designed from a modified jacket to receive heart rate sensors, temperature, oxygenation and positioning. Physiological data are collected and sent to the cloud and after treatment, it generates charts of the practitioner's accompaniment during their running or walking.The system was tested in real environment and presented significant results for the development of a commercial prototype.
- 17:08 Smartphone as Unobtrusive Sensor for Real-time Sleep Recognition
- Sleep is fundamental to health, performance and well-being. Studies demonstrate that, in some countries, sleep disorders are reaching epidemic levels. For this reason, automatic sleep recognition systems can be helpful, on the one hand, to foster self awareness of own habits and, on the other, to implement environment management policies to encourage sleep. In this context, we propose an unobtrusive smartphone application for real-time sleep recognition which relies on contextual and usage information to infer sleep. We tested selected features using kNearest Neighbors, Decision Tree, Random Forest, and Support Vector Machine classifiers. Experimental results demonstrate the effectiveness of the proposed approach, achieving acceptable results in term of Precision, Recall, and F1-score.
- 17:09 Multi-Robot Engagement in Special Education: A Preliminary Study in Autism
- An engagement of humanoid (social) robots in special education may raise ethical questions. Hence, the development of useful therapeutic treatments may be hampered. To circumvent the aforementioned problem we propose training a humanoid robot R1 toward copying (i.e., reproducing) the behavior of a child. Then, use another humanoid robot R2 to interact with robot R1 toward developing an effective treatment in special education without raising ethical questions. Experiments here consider children with autism. Preliminary application results have been encouraging.
- 17:11 Early Detection System of Dementia Based on Home Behaviors and Lifestyle Backgrounds
- The number of dementia illness has been increasing around the world. Dementia disease can be treated if it has been early detected. In previous work, we have developed an early detection system for dementia. Behaviors in the initial stage of dementia are many and diverse from designated to subjects, therefore, in this study, we collected data from different subjects and also we considered lifestyle backgrounds. Therefore, in this research, we improved the sensor device so as not to be influenced by the environment, and made it possible to accumulate data of persons suffered from dementia that had not been accumulated.
- 17:12 A Novel Method to Generate a High-Quality Image by Using a Stereo Camera
- In this paper, we propose a stereo camera system composed of a monochrome camera capturing a high-quality monochrome image and a color camera acquiring color information of a scene. By fusing two images captured from each camera, the proposed stereo camera system generates a high-quality color image. In the proposed method, stereo matching is first performed in which the color image is aligned with the monochrome image. Then, in the aligned color image, inappropriate regions that contain incorrect color information are removed by using the Gaussian mixture model, and regions without any color information are compensated by using the super-pixel algorithm. The resultant image is obtained by transferring the color information of the modified color image to the monochrome image. Experimental results show that the proposed algorithm can produce high-quality color images.
- 17:14 Asian-style Food Intake Pattern Estimation Based on Convolutional Neural Network
- Monitoring food intake behavior is a great help for people who need to manage or prevent obesity. Many researchers have proposed an automatic food intake monitoring technologies based on an image-based method and an inertial sensor-based method and so on. However, most previous works require additional equipment such as a camera on the table, ear clip type sensor with a microphone. Also, previous researchers have not developed on recognition methods considering various food intake cultures such as eastern and western. In this paper, we develop an algorithm that recognizes various food intake patterns, especially Asian cultures with accelerometer sensor without the need for additional sensors outside. We use three axis accelerometer sensors of a wearable device and apply convolutional neural network (CNN) to recognize various kinds of food intake behaviors such as eating with a spoon, picking food with chopsticks, and drinking water. The proposed model was verified by acquiring data from 8 subjects and showed 72.95% accuracy.
- 17:16 A Priority Mutex with Bounded Waiting for Interactive Consumer Electronics Systems
- In most operating systems for interactive consumer electronics devices, inter-process communication (IPC) primitives use mutex locks to maintain consistency against concurrent accesses. However, using a real-time mutex (rt_mutex) in the IPC component can cause indefinitely long starvation of low priority processes. Because such starvation is not suitable for soft real-time systems, we propose a novel mutex mechanism that prioritizes the high priority processes while guaranteeing the bounded waiting of low-priority processes. Our evaluation showed that the waiting time of the low priority processes for IPC was restricted up to 68 ms while the response time of the high priority process was similar to rt-mutex's (13 ms).
- 17:17 Constructing Secure Peer Data Connectivity for Mobile Systems
- Software Constructed Push-Pull data network architectures employ multiple bridged peripheral links to create an ultra-fast, ultra-secure and private data transfer platform. Bridging standard USB 3.0 technologies, we demonstrate a universal and scalable data platform offering the highest level of data privacy, security and performance for mobile systems. Supporting heterogeneous connections across different protocols, data may flow easily across the entirety of available consumer electronics systems with unparalleled speed and flexibility.
- 17:19 Stability Improvement of an Accelerated Android Operating System for Application Observation
- We focus on Android operating system that is based on Linux kernel and a method for accelerating time flow speed recognized by the processes in the system. The method accelerates the speed by modifying the time managing implementation in the kernel and reduce time to monitor applications' behavior. We discuss the stability and accuracy of the method with high acceleration ratios.
- 17:20 Complex Activity Recognition System Based on Cascade Classifiers and Wearable Device Data
- This paper proposes a system for recognizing human complex activities by using unobtrusive sensors such as smartphone, smartwatch and bluetooth beacons. The method encapsulates two classification stages. The former is composed of two parallel processes: the Main Activity Detection (MAD) and the Room Detection (RD). The latter implements the Complex Activity Detection (CAD) process by exploiting the outputs of the first stage and the accelerometer data of the smartwatch. The cascade classification approach that combines the room detection with the main/complex activities recognition task constitutes the novelty of the work. Preliminary results demonstrate the reliability of the system in terms of accuracy and macro-F1 score.
- 17:22 Study of the Smartphone Operation Support System for Visually Impaired People
- In this paper, we propose the smartphone operation support system for visually impaired. It is difficult for visually impaired people to grasp complicated screen configuration of smartphone. Visually impaired people have a problem that the people don't know the operation position of touch panel. Therefore, we created the touch screen application for visually impaired people to get the information easily. As a result, when comparing conventional application and proposed application in three environments "screen display", "screen non-display", and "blindfold", we confirmed a drastic reduction in operation time of proposed method.
- 17:24 Tone Mapping for Video Gaming Applications
- Due to the increasing interest in physical-based rendering in High Dynamic Range (HDR) in game development and the lack of commercial HDR displays, tone mapping the HDR game content to match the capabilities of Standard Dynamic Range (SDR) displays is currently a topic of high importance. In this paper, we propose a global Tone-Mapping Operator (TMO) for video gaming applications which takes advantage of the unique characteristics of the rendered HDR content to reproduce a tone-mapped scene that best matches the appearance of the original HDR scene in terms of preserved global contrast and texture details. This results in a more appealing game and increases players' quality of experience.
- 17:25 Automatic 3D Face Component Analysis Technique
- This paper proposes a technique to perform segmentation for the meaningful regions that part of the face captured by 3D scanners or 3D sensors, automatically. Each part recognition of the scanned face is vital for the 3D applications such as modeling, animation and 3D printing. We transfer the template model labeled with the meaningful part to the scanned face model to find the corresponding part of each meaningful part of the template model. This technique can be used to the several applications such as 3D face modeling, facial animation, virtual facial surgery and 3D printing.
- 17:27 A Virtual Twinning System for User-Centered Eidetic Human-Object Interaction
- Virtual reality (VR) offers a uniquely experience to interact with imaginary items or features by simulating a user's physical presence in a virtual environment. Recently, VR services that achieve a higher level of realism using advanced VR equipment have been attracting public attention. However, studies for interworking physical devices with digital objects in a virtual environment are still insufficient. In this paper, we propose a virtual twinning system which can provide a user-centered eidetic IoT service in a VR environment by linking physical things to virtual objects.
- 17:28 GPU-Accelerated High-Resolution Image Stitching with Better Initial Guess
- High-resolution image processing is crucial for emerging applications in virtual reality. The trade-off between robustness/accuracy and speed pose a challenge to high-resolution image stitching. This paper accelerates the state-of-the-art image stitching algorithm by using the CUDA toolkit. We also use the mesh coordinates as an initial guess for solving warping function for an additional increase in speed. Experimental results show that our implementation achieves comparable quality to the state-of-the-art work, while our implementation is almost 2x faster on high-resolution images (3264x2448).
Sunday, January 14 17:30 - 18:30
- 17:30 Viability analysis of content preparation configurations to deliver 360VR video via MPEG-DASH technology
- We present a viability analysis of a content preparation scheme used to smartly distribute 360VR video using the MPEG-DASH paradigm. Several configurations in which the portion of the scene that is located at the opposite side of the viewport is set to black have been analyzed in terms of bandwidth saving
- 17:45 Design of False Heart Rate Feedback System for Improving Game Experience
- This work investigated effect of false heart rate (HR) feedback on game experience. Our experimental results indicated that false HR feedback not only prevented from getting tired of the game but also enhanced player's experience. Furthermore, a new game controller that can present HR information audibly and tactually was developed.
- 18:00 HLS-based 360 VR using Spatial Segmented Adaptive Streaming
- Recently, by advances in VR (Virtual Reality) contents and HMD (Head Mounted Display), 360VR video related research and development have been actively progressed. Also, mostly recent VR contents are provided with ultra-high definition, over 4K (UHD) and 8K (SUHD). The transmit efficiency which using the most efficient video compression, H.265, to handle such 360VR videos can be effected due to over-transmitting unseen fields in network streaming service. In this paper, a server and a network load problem can be solved by extracting and utilizing information in user-concentrated FOV (Field of View). Regarding to this concept, we propose the Spatial Segmented Adaptive Streaming (SSAS) method. By transmitting original quality video in a currently concentrated field, while transmitting degraded quality video in other fields, network load can be reduced. However, this selectively transmit method has caused switching quality delay by FOV movement. Therefore, we propose the HLS-based real-time adaptive streaming method through video fields and pre-encoding per quality.
- 18:15 An Improved Design of A Lighting Switch System with Augmented Reality Technology
- In the traditional wall-type power switch, there are usually several switches group together, even to six. If there is no note or marker on the switch, the user will be confused because they do not know each switches corresponding to another. In this design, we use the augmented reality technology to reality it. It uses the smart phone, active the built-in camera and according to the special icon on the wall. A virtual and attached switch will pop out, and then the user only presses the icon on-screen to turn on the corresponding switch. Most of the wall-type power switches are used in the lighting system, so in this article that use the on / off light switch as an example.
- 17:30 Scalable Monitoring of End-to-End Delay in Live Video Services
- In this article, we propose a simple method to monitor the end-to-end live lag in a way which is scalable and meaningful for the service providers. The proposal is based on detecting the wall-clock timestamp of the video frames at capture time and inserting it in-band, so that it is available at any point in the network. Live lag can be monitored at any point of the chain, with the same accuracy as the one provided by the clock synchronization mechanism used.
- 17:45 A Study on DRM Application on High-quality Audio
- Due to the rapid increase in the production of portable digital audio devices and the continuous upgrades to internet access speeds, the market for MQS(Mastering Quality Sound) audio is growing rapidly as the format possesses higher audio quality than MP3. Unlike MP3 or CD formats that result in loss of quality and resolution, MQS represents the original audio recording of the singer in the studio with no loss in audio quality, and utilizes lossless audio codecs for compression to maintain the highest quality. While the most commonly used lossless audio codec is FLAC, the compression rate of this codec is not effective, resulting in large file sizes and overhead when conventional DRM is applied. This paper presents a new DRM model that allows for low-quality previews and minimizes battery drain in smartphones. This model has already been implemented on the Android platform, and usability has been evaluated.
- 18:00 Frame Sequential Video System with Complementary Pastel Color and Its Conversion
- We have proposed the frame sequential system as an alternative method which has one color plane in a frame. In this system, it is not easy in the case of high saturation color images. To avoid this problem, we subtract primary color from white. This format is named "Frame Sequential with Complementary Pastel Color". About the motion estimation, Y similar methods are useful for new planes. Furthermore, the anti-error methods to estimate the motion vectors should be enhanced. Finally, the optimal color matrix and the improvement of the motion estimation are discussed.
- 18:15 Computational Complexity Reduction of Two-Channel PMWF Using Frequency Averaging Technique
- This paper proposes a frequency averaging method to reduce computational complexity for the implementation of a two-channel PMWF (parameterized multichannel Wiener filter) algorithm. Experimental results show that the proposed algorithm can reduce around 40 % of computation compared to the conventional PMWF algorithm while preserving speech quality after the speech enhancement.
- 17:30 Reconstruction of Conformal Array Beam Pattern Using Compressive Sensing
- In this paper, compressive sensing is applied to correct the beam pattern of 3-D conformal array. Compared with the conventional 3-D interpolation technique, it shows excellent characteristics.
- 18:00 RCS Prediction Using Cauchy Method with Increasing Prediction Frequency
- In this paper, we apply the Cauchy method, the LPF, the SVD and the TLS theory using the RCS data in the low frequency band and predict the RCS data in the high frequency band. As the predicted frequency increases, the error between the original data and the predicted data is confirmed.
- 17:30 A Novel DC Circuit Breaker Design Using A Magnetically Coupled-Inductor for DC Applications
- Attempting to simplify the steps of converting power into its simplest form, new microgrids will be conceptualized in the DC format. Certain power systems and power conversion components are highly available but, as it applies to DC circuit breakers, various designs are currently in the experimental stage. The most significant drawback of such a system is that interrupting current that is without zero crossing causes a sustained power arc. This paper introduces a novel circuit breaker model for DC, which utilizes a coupled inductor circuit breaker, and a mathematical model of insulated gate bipolar transistor (IGBT) has been developed.
- 17:45 Lab Automation Drones for Mobile Manipulation in High Throughput Systems
- In this paper, a lab automation drone notional concept is introduced. Here, a robotic limb is attached to a robotic rotorcraft. The limb's gripper allows the unmanned aerial vehicle to dexterously manipulate objects such as micro-arrays and test tubes often used in high throughput systems (HTS). The resulting drone could augment existing HTS operations. The 6 degree-of-freedom (DOF) arm and gripper design are presented. Test-and-evaluation approach and results are also given.
- 18:00 Detecting Faulty Solar Panels Based on Thermal Image Processing
- We propose a faulty solar panel detecting system based on thermal image processing.
- 18:15 Solid State Cache Management Scheme for Improving I/O Performance of Hard Disk Drive
- In this paper, we propose the Hybrid Unit 2Level-LRU scheme to optimize the slow I/O speed of the HDD. We minimized disk access and increased the cache hit rate by frequently accessing the pages in the SSC. In addition, we minimized disk head movement by maintaining the adjacent pages requested to write in a group when the pages are destaged to the HDD. Depending on the read requests workload, the cache hit ratio can be improved if there are more pages frequently accessed to read than to write. We will describe these techniques and discuss the results of experiments.
- 17:30 Flying Animals and the Art of Presentation
- This is an informative and entertaining session for Young Professionals (Students and higher grade members with short career) This session will start with an expert's giving tips about how to make an impactful presentation. After this talk, several speakers give a short presentation of paper for the 2018 ICCE conference. Their presentation will be judged by the audience who will have an opportunity to express their opinions using flying toy animals!