Program for 2017 IEEE International Conference on Consumer Electronics-Asia (ICCE-Asia)

Time Ball Room 1 Ball Room 2 Break And Lunch Area ICCE Tutorial Room 1 Session

Thursday, October 5

09:30 am-10:00 am         Tutorial Registration
10:00 am-01:00 pm       Deep Dive into Tensor Flow  
02:00 pm-05:00 pm       Vulkan - Next Generation Graphics and Compute  

Friday, October 6

08:30 am-09:15 am     Registration    
09:15 am-09:30 am Conference inauguration        
09:30 am-10:00 am Inaugural Keynote - IoT & Smart Audio: a Glimpse of CE Industry Trend        
10:00 am-11:00 am Industry Expert Keynote - Enabling Mobile eXtended Reality        
11:00 am-11:30 am     Tea and Networking Break    
11:30 am-01:00 pm Session 1a : IoT and Cloud Computing Session 1b : Artificial Intelligence and Machine learning for CE Applications   Session 1c: Emerging Technology  
01:00 pm-05:15 pm   Session 2: Startup Product Demonstrations / Poster Presentation Lunch    
02:00 pm-02:45 pm Industry Expert Keynote: Commercializing Hardware Innovations      
02:45 pm-03:15 pm Product Presentation      
03:15 pm-03:45 pm   Tea and Network Break    
03:45 pm-04:00 pm WIE Track - Introduction and Welcome      
04:00 pm-04:30 pm WIE Track - Apache Spark on Hadoop - Optimisation and Parallelisation      
04:30 pm-05:00 pm WIE Track - Software Defined Networks      
05:15 pm-06:15 pm Banquet Keynote - "What I Learned from Living in Future"        
07:00 pm-09:30 pm Banquet Dinner        

Saturday, October 7

08:30 am-09:00 am     Registration    
09:00 am-10:00 am Keynote - Human motion detection and use in Entertainment/Gaming using IMUs, Integrated sensor processing systems - Challenges, Algorithms        
10:00 am-10:15 am     Tea and Network Break    
10:15 am-12:00 pm Session 3a: Artificial Intelligence and Machine learning for CE Applications Session 3b Keynote: Disruptive Technologies in the Automotive Industry   Session 3c : Energy Management and Semi Conductor Devices  
10:45 am-12:00 pm Session 3b Papers: Automotive CE Applications    
12:00 pm-01:00 pm     Lunch    
01:00 pm-02:00 pm Keynote - Creating Smarter Machines through the Power of Analytics        
02:00 pm-03:45 pm Session 4a: Robotics Technologies Session 4b Keynote: Brain Computer Interfaces   Session 4c: Automotive CE applications  
02:45 pm-03:45 pm Session 4b Papers: Brain Computer Interfaces    
03:45 pm-04:00 pm     Tea and Network Break    
04:00 pm-04:45 pm Panel Discussion: "Innovation in the new age - Challenges, directions"        
04:45 pm-05:15 pm Valedictory Function and Awards Ceremony        

Thursday, October 5

Thursday, October 5, 09:30 - 10:00

Tutorial Registrationgo to top

Thursday, October 5, 10:00 - 13:00

Deep Dive into Tensor Flowgo to top

Manoranjan Padhy, Google
Room: ICCE Tutorial room 1

Thursday, October 5, 14:00 - 17:00

Vulkan - Next Generation Graphics and Computego to top
Room: ICCE Tutorial room 1

Friday, October 6

Friday, October 6, 08:30 - 09:15

Registrationgo to top

Room: Break and Lunch Area

Friday, October 6, 09:15 - 09:30

Conference inaugurationgo to top

Ball Room 1

Friday, October 6, 09:30 - 10:00

Inaugural Keynote - IoT & Smart Audio: a Glimpse of CE Industry Trendgo to top

Dr Sharon Peng, President, IEEE CE society
Ball Room 1

Friday, October 6, 10:00 - 11:00

Industry Expert Keynote - Enabling Mobile eXtended Realitygo to top

Dr Ajit Rao, Qualcomm Technologies
Ball Room 1

Friday, October 6, 11:00 - 11:30

Tea and Networking Breakgo to top

Room: Break and Lunch Area

Friday, October 6, 11:30 - 13:00

Session 1a : IoT and Cloud Computinggo to top

Ball Room 1
Chair: Nippun Kumaar Aa
11:30 A Review of Big Data Analytics over Cloud
As the volume of data is increasing exponentially, there is a need for better management of data to research and industry. This data, known as Big Data, is now used by various organizations to extract valuable information which can be analyzed computationally to reveal patterns, trends and associations revealing the human interaction and behavior for making various industrial decisions. Due to the large volume of data, it is stored in the cloud and all the analysis is done over Big Data over cloud since it is not possible for traditional systems to store such large amount of data. But the data must be optimized, integrated, secured and visualized to make any effective decision. Analysing of the large volume of data is not beneficial always unless it is analyzed properly. The existing techniques are insufficient to analyze the Big Data and identify the frequent services accessed by the cloud users. Various services can be integrated to provide a better environment to work in. Using these services, people become widely vulnerable to exposure. That is, it becomes possible to collect more data than it is required which may lead to data leakage and hence security concerns. Results can be analyzed in a better way by visuals like graphs, charts etc. and it helps in faster decision making. It also includes MapReduce Algorithm which will help in maintaining a log of users activities in the cloud and show the frequently used services. This paper proposes a scheme for making Big Data Analytics more accurate, efficient and beneficial.
11:52 Securing Network Controlled Systems: Modeling Attacks and Defenses
Networked control systems (NCS) have emerged as a viable solution to efficiently manage modern critical infrastructure in manufacturing industries and urban centers. Failure to secure the network exposes the system to attacks from malicious entities that could trigger serious economic losses or endanger plant safety. The shared communication network that interconnects the spatially distribute control components coupled with safety critical nature of data necessitates the need to secure NCS. In this work, we propose a mechanism to secure NCS against denial of service (DoS) and deception attacks. Evaluation in MATLAB/Simulink environment demonstrates the feasibility and effectiveness of the proposed approach in securing NCS
12:15 Towards a Smarter Power Grid- An IoT and Data Analytics Based Solution
Smart Grid and Internet of Things (IoT) are the latest trends respectively in the domains of Power Engineering and Computer Science. The amalgamation of the two presents enormous potential to improve the efficiency, security and reliability of the power grid. In this paper, the authors review existing solutions and propose their ideas, as an improvement over existing techniques to realize the idea of a Smart Grid and a competitive Energy market by application of Mathematical and Statistical tools to analyze and predict load demand as well as agent behavior in the present scenario.
12:37 Detection of Transformed Malwares using Permission Flow Graphs
With growing popularity of Android, it's attack surface has also increased. Prevalence of third party android marketplaces gives attackers an opportunity to plant their malicious apps in the mobile eco-system. To evade signature based detection, attackers often transform their malware, say by introducting code level changes. In this paper we propose a lightweight static Permission Flow Graph (PFG) based approach to detect malware even when they have been transformed (obfuscated). Transformations often result in new permissions, thereby changing structure of PFG. To combat such a threat scenario our approach involves comparing PFG of transformed malware with PFG of non-transformed malware by leveraging graph similarity algorithm. We propose graph based similarity metrics at both edge level and vertex level in order to bring forth the structural similarity of the two PFGs being compared. We validate our proposed methodology through machine learning algorithms. Results prove that our approach is successfully able to group together Android malware and its variants (transformations) together in the same cluster. Further, we demonstrate that our proposed approach is able to detect transformed malware with a detection accuracy of 98.26%, thereby ensuring that malicious app can be detected even in presence of transformations.

Session 1b : Artificial Intelligence and Machine learning for CE Applicationsgo to top

Ball Room 2
Chair: Sudarshan Tsb
11:30 Efficient Frequency Domain CNN Algorithm
Deep Learning techniques like Convolutional Neural Networks (CNN) are getting popular for image classification with broad usage spanning across automotive, industrial, medicine, robotics etc. Typical CNN network consists of multiple layers of 2D convolutions, non-linearity, spatial pooling and fully connected layer, with 2D convolutions constituting more than 90% of total computations. The Fast Fourier Transform (FFT) based approach for convolution is promising in theory, but not used in practice due to growth in memory sizing of coefficients storage. The paper proposes new frequency domain algorithm which avoids memory size growth compared to traditional FFT based approach for performing 2D convolution. The proposed algorithm performs Fourier Transform (FT) of coefficients On-The-Fly (OTF) instead of offline calculation on PC. The proposed algorithm consists of expands, OTF-FT and pruning blocks that do efficient 2D convolution in the frequency domain. The proposed algorithm is compared with the FFT-based algorithm for the coefficient transformation. As per simulations, assuming typical network configuration parameters, the proposed algorithm is 4-8X faster compared to FFT based approach for the co-efficient transform.
11:52 Evaluation of Relevance Vector Machine Classifier for a Real-time Face Recognition System
Face recognition has found a variety of applications in consumer electronics, such as laptops, smart phones, home security systems, home automation systems and many more. Machine learning is one of the important concepts, required for designing any pattern recognition system, including the proposed real-time face recognition system. Relevance vector machine is considered as one of the most recent machine learning algorithms reported in literature. In this paper, design and evaluation of Relevance Vector Machine classifier architectures for a real-time face recognition system using Histogram of Oriented Gradient features is proposed. In order to assess the performance of system designed, AT&T database of faces are initially considered, followed by the performance evaluation using real-time face inputs from the system camera. 81.25 − 97.00% recognition accuracy is obtained on using the proposed system and the proposed work can be easily extended for various other pattern recognition systems too.
12:15 Design of a Phoneme Based Voice Controlled Home Automation System
Speech recognition has found a variety of applications in consumer electronics, such as smart phones, laptops, speech to text converters, live subtitling on television, dictation tool and many more. In this paper, design of a Voice Controlled Home Automation System using phoneme based speech recognition is proposed. The proposed system uses Mel Frequency Cepstral Coefficients for feature extraction and a two stage Support Vector Machine for classification. In order to assess the performance of system designed, voice samples were recorded in the laboratory to train the system. 86.00 - 100.0% recognition accuracy is obtained on using the proposed system with a recognition time of 0.9063s. The proposed work can be easily extended for various other voice controlled pattern recognition applications.
12:37 Electrooculography Based Assistive Technology for ALS Patients
This paper describes an Assistive technology for ALS patients that is controlled using Electrooculogram signals (EOG). In situations of traumatic diseases or paralyzed conditions, the patient may lose the ability to communicate with others either by speech or gestures. In most cases, the only remaining and active capability for patients suffering from neurodegenerative disorders are the eye movements. The human-computer interface provided is controlled by the eye movements of the patients. Using various eye movements the individual can control the interface in order to communicate with others or request the help of caretaker in cases of need and emergency. The user interface and virtual keyboard provided may be used to examine the performance and usability of the proposed assistive technology.

Session 1c: Emerging Technologygo to top

Room: ICCE Tutorial room 1
Chair: Kousik Ramasubramaniam Sankar
11:30 Auto White Balance Using Dynamic Histogram Matching for AMOLED Panels
This paper presents a novel method of Auto White Balance (AWB) using Dynamic Histogram Matching (AWB-DHM) for AMOLED panels. Existing methods of AWB based on histogram matching always chooses a specific channel based on pre-defined criteria such as channel having maximum cumulative intensity among the R, G and B channels etc.,. Then match the histograms of other two channels with the chosen channel. Since the channel selection is always constant, for a given algorithm, when AWB is applied for images under varying illumination conditions they are prone to color distortion. On the contrary, the proposed approach in this paper analyzes the histogram distribution for each color channel (R, G and B) of an input image and then dynamically decides the best fit channel for histogram matching. Hence, this approach eliminates color distortion in images under all lighting conditions and enhances the visual quality by increasing the Overlap Area (OA). This method increases the Overlap Area (OA) of three channels by at least 18% and contributes to additional power savings of 11% especially for AMOLED panels when the best fit channel chosen has lower cumulative intensity. Moreover, the proposed method is computationally efficient compared to existing methods.
11:48 Solid State Relay Based Inrush Current Limiter with Short Circuit and Under Voltage Protection for Dc-Dc Converters
This paper discusses about a new P-Channel MOSFET based solid state relay developed for limiting the plug-in inrush current in dc-dc converters operating on high input voltage power bus in satellites. This circuit can be used as a Solid State Relay in any on-board systems for power bus on/off operation. It provides plug-in inrush current limiting and also provides the protection against input bus under voltage and short mode failure of components in input portion of dc-dc converter. The main challenge for the proposed design is to use of minimum number of components in the design and miniaturizing the circuit. The proposed design in hybridized form can be used as a 'component' for inrush current limiting inside every converter OR system and this avoids usage of extra components like electro-mechanical relay and related driver circuit. Proposed design was simulated in Pspice and a lab model was wired and tested successfully for 70V power bus operation for powers up to 160W. The test results of the proposed circuit are also presented in the paper.
12:06 Behavioural Study of Memory Allocators for Android Platform
Speed is important for a malloc implementation because if malloc is not fast enough, application writers are inclined to write their own custom free lists on top of malloc. This can lead to extra complexity, and more memory usage unless the application writer is very careful to appropriately size the free lists and scavenge idle objects out of the free list. The objective of this paper is to do behavioural study and analysis of the memory data and access time with different types of malloc (nedmalloc, jemalloc, tcmalloc, musl-malloc etc.) implementations on Android operating system. Analysis results - enabling Android application developers to optimize the applications by choosing the right malloc implementation in alignment of different use cases.
12:24 Experimental Investigation of the Effect of the Attached Inlet and Outlet Pipe Size on a Hall Effect Flow Sensor Frequency
Hall effect flow sensor (HFS), especially YF-S201, is a consumer electronic device which is an inexpensive, reliable and easy-to use flow sensor for measurement of the volume of fluid (mostly liquid) that passed through the cross-section of the sensor. It is generally used in conjunction with Arduino microcontroller in the flow measurement. It has been widely used for accurate measurement of liquid flowrate for industrial and agricultural purposes. In Hall effect sensor, whenever water flows through the sensor, the pinwheel rotor rotates producing a magnetic field. Its speed changes with different rate of flow. The magnetic field is then converted into an electrical pulse by the Hall effect sensor. The rate of flow is thus directly proportional to the pulse output. The flowmeter can then be calibrated to high accuracy level through the pulse rate. This implies that the Arduino-HFS could be an accurate and reliable flow meter if calibration is done properly. However it is not a precision sensor and it has to be carefully calibrated for taking accurate measurements. There are certain factors that can hamper the pulse rate of the Hall effect sensor like the pressure, flow rate and the sensor orientation. The flow rate alone is under the influence of certain factors like the pressure of the fluid flow, the size of pipe used etc. In our present work, effect of the attached inlet and outlet pipe size on a Hall Effect flow sensor frequency is studied. The sensor is used in conjunction with microcontroller Arduino Uno for measuring the frequency of the sensor. Experiments were also done to test the validity of the calibration factor of 7.5 pulses per second, given by most online sources. Our experimental data shows that as the size of the pipe decreases, the frequency increases, that is, the size of the pipe is inversely proportional to the frequency. The flow rate on the other hand is found to be directly proportional to the size of pipe used: larger the size of pipe, larger is the flow rate. Measurements were taken 2-3 times for each pipe for accurate measurements. The accuracy of hall effect sensor is found to be within +3%.
12:42 Centralized Server Based ATM Security System with Statistical Vulnerability Prediction Capability
This paper proposes an architecture to develop an ATM security system with a centralized private server for monitoring all the ATMs in a city. Security related information is captured using a wide variety of sensors and switches and this information is processed and sent to a central server for further analysis and data presentation. Information from the sensors is sent continuously in real time to build a database. The system performs statistical vulnerability analysis using the data received from various sensors from different ATMs and assigns a vulnerability quotient to all the ATMs. Thus, helps to be a central system for identifying security gaps in a particular ATM. This system seeks to be a proactive measure to prevent risks associated with ATMs.

Friday, October 6, 13:00 - 17:15

Session 2: Startup Product Demonstrations / Poster Presentationgo to top

Ball Room 2

Friday, October 6, 13:00 - 14:00

Lunchgo to top

Room: Break and Lunch Area

Friday, October 6, 14:00 - 14:45

Industry Expert Keynote: Commercializing Hardware Innovationsgo to top

Avinash Kaushik, CEO and Founder, Revvx
Ball Room 1

Friday, October 6, 14:45 - 15:15

Product Presentationgo to top

Ball Room 1

Friday, October 6, 15:15 - 15:45

Tea and Network Breakgo to top

Room: Break and Lunch Area

Friday, October 6, 15:45 - 16:00

WIE Track - Introduction and Welcomego to top

Prof Sujatha, BMS CE
Ball Room 1

Friday, October 6, 16:00 - 16:30

WIE Track - Apache Spark on Hadoop - Optimisation and Parallelisationgo to top

Ms Sai Geetha, TESCO Labs
Ball Room 1

Friday, October 6, 16:30 - 17:00

WIE Track - Software Defined Networksgo to top

Ms Pallavi Srinivasa, Director, Cisco
Ball Room 1

Friday, October 6, 17:15 - 18:15

Banquet Keynote - "What I Learned from Living in Future"go to top

Bob Frankston, IEEE CE Society, USA
Ball Room 1

Friday, October 6, 19:00 - 21:30

Banquet Dinnergo to top

Ball Room 1

Saturday, October 7

Saturday, October 7, 08:30 - 09:00

Registrationgo to top

Room: Break and Lunch Area

Saturday, October 7, 09:00 - 10:00

Keynote - Human motion detection and use in Entertainment/Gaming using IMUs, Integrated sensor processing systems - Challenges, Algorithmsgo to top

Prof. Varadhan SKM, IIT Madras, India
Ball Room 1

Saturday, October 7, 10:00 - 10:15

Tea and Network Breakgo to top

Room: Break and Lunch Area

Saturday, October 7, 10:15 - 12:00

Session 3a: Artificial Intelligence and Machine learning for CE Applicationsgo to top

Ball Room 1
Chair: Sudarshan Tsb
10:15 FPGA Implementation of Blind Source Separation Using a Novel ICA Algorithm
Independent Component Analysis (ICA) is considered as a powerful technique to separate individual sources, from a linear mixture of input sources. This has motivated the use of ICA for various consumer and medical applications. A novel Jacobian Matrix rotation based ICA for blind source separation is proposed in this paper. Hardware implementation of proposed ICA algorithm on a Xilinx Virtex-5 FPGA using System Generator Blocksets is proposed and the performance results are reported.
10:41 An Approach Towards Novel Video Steganography for Consumer Electronics
Videos have proven themselves to be the best medium to demand consumer attention. This has motivated the sharing of videos on Whatsapp, Facebook and other social media. Consumer electronics such as mobile phones, tablets and laptops are mainly used for video sharing. In this paper, a novel attempt is made to conceal a secret video in a cover video using the concepts of steganography and discrete wavelet transform, thus achieving the best of both worlds. Also, the proposed work is extended towards concealing two secret videos in a single cover video. The performance evaluation of proposed technique is reported in the paper. It is observed that the reconstructed videos have PSNR ranging from 74.30 - 90.12 dB and the PSNR of stego videos ranges from 68.66 - 70.80 dB. It is also observed that both, stego videos and reconstructed secret videos are visually indistinguishable from the original video.
11:07 Novel OpenVX Implementation for Heterogeneous Multi-Core Systems
Heterogeneous multi-core systems (CPU, GPU, HWA, DSP) are becoming the de-facto norm for multiple computer vision applications across automotive, robotics, AR/VR, and industrial machine vision. This creates a need for a software framework which realizes high utilization of computing elements, low latency, real-time operation and ease of use. For specific applications, multiple proprietary solutions are offered to satisfy few of the above requirements. This paper proposes a solution based on the standard OpenVX specification to address heterogeneous systems. It introduces novel techniques of distributed graph execution across heterogeneous cores, data tiling to address diverse memory constraints and easy to use high-level graph description to describe the application. This novel solution is implemented on TI's TDA family of SoC for mono camera vision application with platform code generated from high-level graph description. The profiling confirms real time operation, low latency by reducing host CPU interaction and achieving 99% utilization across heterogeneous cores.
11:33 Augmented Reality in Broadcasting
In the past few years, Augmented Reality (AR) is an emerging technology in different fields. It integrates virtual with real world and display the object in Three Dimensional view. It gives Three Dimension(3D) view to an object. Introducing AR to Advertisement industry which will change the future of product development companies and manufacturers. We are proposing AR as the future of Advertising, an advanced technology-oriented marketing. Our idea is to integrate this technology with broadcasting which will in turn boost the Marketing strategy to reach the customer effectively and efficiently

Saturday, October 7, 10:15 - 10:45

Session 3b Keynote: Disruptive Technologies in the Automotive Industrygo to top

Ramesh Babu, DGM, Cyient
Ball Room 2

Saturday, October 7, 10:15 - 12:00

Session 3c : Energy Management and Semi Conductor Devicesgo to top

Room: ICCE Tutorial room 1
Chair: Prabindh Sundareson
10:15 Leakage Optimization of Thick Oxide IO/ESD Transistors in 40nm Global Foundry Process
Thick Oxide ESD transistor in 40nm Global Foundry process is studied for reducing leakage while being area efficient and maintaining performance. Gate induced drain leakage(GIDL) and source-drain leakage were found to be the major leakage contributors. Optimum architecture and sizing are found for IO/ESD design and presented in this paper.
10:36 Sense Amplifier Based High Speed Flip-Flop Design for Advanced Sub-Micron FinFET Standard Cell Library
A novel high-speed sense-amplifier based flip-flop is presented in this paper. The Proposed flip-flop design has improved setup time, D2Q delay and glitch-less output. A Comparative analysis of sense amplifier's stacking order is discussed and the Proposed Sense Amplifier designs exploit multiple stacking orders of the clocked devices, input and feedback transistor and show approximately 11\% overall advantage in performance of the flop as compared to the conventional variant of a flip-flop. Simulation results were obtained for sub-micron FinFET technology using production characterization setup. Monte Carlo stress checks were performed on the proposed designs to ensure bug-free operation and high-yield at the time of fabrication.
10:57 Robust Event Trigger Generation for Post Silicon Validation
It is extremely difficult to eliminate all the design flaws that may exist in a complex integrated circuit design before fabrication due to limitations of pre-silicon verification methodology. Post-silicon validation has become pivotal in order to fix these design flaws which could not be identified during pre-silicon verification. One of the most effective silicon debug techniques is tracing internal signals during circuit's normal operation and using the gathered information to identify and fix the root cause of the problem. Trace-based technique involves logic analysis that works on the principle of event triggering. Existing trigger generation methods suffer from the false multi-triggering problem due to dependency on exact value of delay provided by internal delay elements and do not give a stable performance against PVT variations. This paper demonstrates a robust technique for generating these triggers which is independent of exact value of delay provided by internal delay elements and immune to false multi-triggering along with a consistent behavior across the PVT variations.
11:18 Power Mitigation in High-Performance 32-bit MIPS-based CPU on Xilinx FPGAs
The proposed work is a design of a 32-bit MIPS-based CPU with a high performance five stage pipeline, to incorporate power optimization techniques for the processor. The functionality of design is verified by writing Verilog Modules on Xilinx 14.5 choosing the target FPGA device. The design helps to improve the speed and increase the throughput of the processor. Simulation Results has been carried out from ModelSim 6.2c. Analyzing the design floorplan, study the detailed netlist and implementation was done on PlanAhead tool gives accurate results. From the performance viewpoint, FPGA based implementation of processor totally concentrated on designing the architectures in HDL with the aim of increasing the overall speedup of CPU and reduce the power consumption at 45nm technology. The salient features of this work are increased the number of instructions, high speed and minimize the power consumption with low power HDL modification techniques. The design has consumed power less than 119mW with the maximum frequency of operation 70.413MHz. Optimized power observed about 22.72% after applying power reduction techniques which make this work useful for low power consumption in FPGAs.
11:39 Current Controlled Capacitor Less Low Dropout Voltage Regulator for Fast Transient Response
This paper discusses and address a new current control transistor based low dropout (LDO) regulator offering a high open loop gain and large bandwidth without using complex compensation and large output capacitor. The conventional error amplifier based Ldo which is using voltage feedback loop suffers from an inherent poor load transient response. The poor load transient associated with error amplifier results because of limited DC open-loop gain and limited closed loop bandwidth. The frequency response of error amplifier based Ldo is highly sensitive to load current. Current controlled transistor based LDO has been designed for fast dynamic response with low output capacitor in the range of 1 to 10 uF. Experimental results show circuit has small output voltage variations of less than 10 mV for load current variation in the range of 0-3A without any overshoot or undershoot. This circuit is also simulated in PSpice and results are compared with Hardware LDO.

Saturday, October 7, 10:45 - 12:00

Session 3b Papers: Automotive CE Applicationsgo to top

Ball Room 2
Chair: Venkobarao Vivek
10:45 Adverse Weather Simulation for Training Neural Networks
Convolutional neural networks generally require considerable amount of data for training to perform adequately well in all real-world scenarios. Many times, the data for all scenarios is hard to collect and ground truth annotation is also a challenge. Similar problem exists in training networks for the autonomous vehicles given the diverse weather conditions where these cars are expected to be driven. Thus, a synthetic data generation model is imperative and we go about building a weather simulation framework. This framework is intended to generate weather conditions over different driving scenarios. To start with, we go about implementing a completely configurable rain/fog/windshield simulation model. The scope of this framework, however is much more than these two models. Apart from refining these models further as and when need, we intend to build in mechanisms to simulate more diverse weather conditions within this framework.
11:10 Flexible and Efficient Perspective Transform Engine
Perspective transform (or Homography) is commonly used in image and computer vision processing, which typically solved using a back mapping algorithm. It is used in multiple use-cases e.g. view point change, fish-eye lens distortion correction, chromatic aberration correction, stereo image pair alignments etc. Prior work in literature addresses individual use-case by fixing data format, interpolation method and block sizes due to constraints of irregular data fetch. In this paper, we propose an area efficient and flexible HW IP solution for perspective transform engine to address multiple use-cases with software controlled data format, interpolation method and block size. The proposed solution consists of multiple innovative techniques namely On-The-Fly (OTF) transform, Off-line mesh based Look up Table (LUT) with optional down-sampling, corner pre-fetch and dual interpolation. The proposed solution achieves clock up-to 300 MHz with output pixel throughput of 1 cycle/pixel and 0.21 mm2 of silicon area in 28nm LP-CMOS process node
11:35 Parallel Image Pre-Processing for In-Game Object Classification
Games that involve photo-realistic rendering of virtual worlds, Special effects, VR, involve highly complex calculations on the GPU (Graphics Processing Unit). While GPUs have specialized cores that accelerate Graphics operations, they are also capable of general purpose computing. In this paper, a specific data flow is chosen for the use-case of in-game object-classification. This use-case involves converting large input graphics resolutions (4k) to much lower resolutions (256x256) needed for compute. We compare the different approaches available for executing the data flow using CUDA (Compute Unified Device Architecture) without impacting gaming performance, and publish comparisons on different classes of GPUs for the first time. It is shown that best performance is achieved with a combination of well optimized algorithms, priority of work assignment, and a flexible execution framework.

Saturday, October 7, 12:00 - 13:00

Lunchgo to top

Room: Break and Lunch Area

Saturday, October 7, 13:00 - 14:00

Keynote - Creating Smarter Machines through the Power of Analyticsgo to top

Suman Narayan, Senior VP, Cyient
Ball Room 1

Saturday, October 7, 14:00 - 15:45

Session 4a: Robotics Technologiesgo to top

Ball Room 1
Chair: Sudarshan Tsb
2:00 Continuous, Robust Hand Gesture Recognition for Embedded Devices
Gestures are gaining a significant demand in embedded devices to provide enhanced user experience through natural communication. This paper presents a robust approach of Hand Gesture Recognition to detect gestures in real time without the need of a delimiter. The method represents the gestures with trajectory strings and matches them with stored trajectory strings of pre-defined gestures. The proposed method employs an enhanced matching scheme using a custom edit distance metric that enables to find the closest gesture in the database with speed invariance (no need to control the speed of the gesture), position invariance (can start the gesture from anywhere in space) and scale invariance (can be far away or close to the camera while performing the gesture) without the need of a delimiter (ability to start at any time). All these above mentioned characteristics are ideal for any gesture recognition system and our method provides robustness w.r.t all these and is also pretty inexpensive in terms of computation while not compromising much with accuracy, thus making it very suitable for devices with weak to moderate processors including most of the embedded devices. We have tried fairly complicated circular gestures and the results were promising.
2:21 Design, Analysis and Control of an Autonomous Underwater Surveillance Robot
A large part of our oceans is still an uncharted territory to humans. The coastal and river bodies serve as a channel for transportation and trades, and hence are vulnerable to accidents and mishaps. In some unfortunate case of an airplane crash, it becomes really hard to find an object underwater, and it also put human lives at stake. Thus, a continuous probation is needed. The aim of this paper is to study the design and development of an Autonomous Underwater Surveillance Robot which would propel this human endeavor and can also be deployed for security surveillance for the Navy. The design and development of the robot took place in a controlled environment. The paper aims to suggest a design for the robot with functionalities for maneuverings, image recognition and depth control. The paper revolves around designing the robot and performing various simulation analysis so as to get a clear picture of its condition in the real life simulation. It needs intelligent platforms to solve such complex problems. Using various software packages such as Solid Works and OpenProp for the design and analysis part and FlowExpress for pressure and Computational Fluid Dynamics (CFD) simulation. The proposed model for the robot is a robust design with concern to all the surrounding and the environmental conditions. The result is a streamlined design for the AUV facing the least underwater drag.
2:42 Towards a Scalable Hardware/Software Co-Design Platform for Real-time Pedestrian Tracking Based on a ZYNQ-7000 Device
Currently, most designers face a daunting task to research different design flows and learn the intricacies of specific software from various manufacturers in hardware/software co-design. An urgent need of creating a scalable hardware/software co-design platform has become a key strategic element for developing hardware/software integrated systems. In this paper, we propose a new design flow for building a scalable co-design platform on FPGA-based system-on-chip. We employ an integrated approach to implement a histogram oriented gradients (HOG) and a support vector machine (SVM) classification on a programmable device for pedestrian tracking. Not only was hardware resource analysis reported, but the precision and success rates of pedestrian tracking on nine open access image data sets are also analysed. Finally, our proposed design flow can be used for any real-time image processing-related products on programmable ZYNQ-based embedded systems, which benefits from a reduced design time and provide a scalable solution for embedded image processing products.
3:03 Intelligent Monitoring and Maintenance of Solar Plants Using Real-time Data Analysis
In the recent years, we have seen a rapid increase in the worldwide installations of solar plants. A solar plant consists of various different types of devices (strings of solar panels, inverters, transformers etc.). The overall energy generation and performance of a solar plants depends upon effective and timely maintenance of such different devices. These devices can degrade over time or due to specific malfunctioning in the equipment. Energy production is not only affected by internal factors at plant but by the external factors like irradiation and module temperature as well. In this paper, we present a system to identify various malfunctioning / possible break-downs of such devices based on real-time monitoring and various real-time anomaly detection techniques. Once the anomaly is identified, it is immediately informed to the appropriate service engineers for timely action. It helps in providing increased plant performance and efficiency for the solar plants.
3:24 Swarm Home Robots
Swarm robotics is a relatively new research area inspired from biological systems such as ant or bee colonies. It composes a system consisting of many small robots with simple control mechanisms capable of achieving complex collective behaviors on the swarm level such as aggregation, pattern formation and collective transportation to name a few. The main motto of this paper is to incorporate the swarm technology in consumer electronics where these are used in the operations like pick and place, transportation of the house-hold items from one location to another location in the place we reside. While a single robot can perform the home tasks, it might not be able to perform the tasks quickly. Time will be lost in finishing one task and then moving to another. So, it is evident that more than one robot must be used here. When more than one robot is used, controlling the robots becomes an important issue especially when the target locations and the robots are large in number. So, for easing the control of more than one robot, traits of Swarm behavior are replicated and implemented. Hence, a swarm behavior replicating group consisting of two robots are designed and fabricated that demonstrate how multiple-tasks can be handled effectively by a group of robots that perform individual actions while communicating and acting symbiotically. In the present paper, when target co-ordinates are passed to one robot, while retaining one target location, it passes another location to the other robot. Both start moving towards their target locations, but continue communicating with each other. In event of failure of one of the robots, the other robot takes notice of the failure. It continues to its location, does its work at that location, and then gives a backup at the other target location by performing the task of the robot that failed.

Saturday, October 7, 14:00 - 14:45

Session 4b Keynote: Brain Computer Interfacesgo to top

Srichandra, Sr Director Standards & Technology
Ball Room 2

Saturday, October 7, 14:00 - 15:45

Session 4c: Automotive CE applicationsgo to top

Room: ICCE Tutorial room 1
Chair: Venkobarao Vivek
2:00 A Novel Thermocouple for Ultra High Temperature Applications: Design and Computational Analysis
The development of multi-purpose sensors for sensing heat flux and temperature has become critical for operation of high temperature structures viz. gas turbines, furnaces and hypersonic space vehicles. Conventional thermocouples have limitations owing to the maximum operational temperature of metallic alloys. In this perspective, it is imperative to explore new material systems which can extend the operational range of thermocouples up to 2500K temperature. Current work investigates a novel thermocouple design which employs Zirconium diboride based ultra-high temperature ceramics for providing protection to metallic thermocouple wires in oxidizing environments in heat flux as high as 2.5MW/m2 for up to 125s duration. The performance evaluation of this novel thermocouple design has been carried using Finite element based computational modelling. This demonstrates that the proposed thermocouple design has good sensitivity of 3µV/K in relevant hot environment.
2:35 Novel Architecture for Cloud Based Gen Next Vehicle Platform - Transition from Today to 2025
The sole idea of the autonomous driving is to reduce the number of accidents which will be caused by the human errors. There will be a phase where autonomous vehicles will have to take part with the human driven vehicles. In this phase autonomous vehicle and vehicle drivers must be careful in driving maneuvers, since it is impossible to just have autonomous vehicles on the road while the transition takes place.
3:10 Throttle Override Safety System for 'Design Induced Pilot Error' in Electric Vehicles
Electric or Battery driven vehicles are the present and future of road transport. Safety systems for these vehicles are important due to the powertrain dealing with high current and absence of clutch. This paper provides solution to a specific problem called 'design induced pilot error'. Such errors occur due to placement of pedals in the vehicle or due to lack of experience and as a result, both the throttle and brake pedals are pressed simultaneously. These incidents may have serious repercussions, such as toppling of the vehicle or other such accidents. It is solved by overriding the throttle pedal when incidents like 'panic braking', 'hard braking' occur. The circuit, realized using analog circuitry detaches the drive train from the shaft, thus freewheeling the vehicle. This leads to only the brakes interacting with the wheels, thus reducing the risk of a calamity. This system, in addition to aiding inexperienced drivers, provides better control to experienced drivers in maneuvering the vehicle.

Saturday, October 7, 14:45 - 15:45

Session 4b Papers: Brain Computer Interfacesgo to top

Ball Room 2
Chair: Kousik Ramasubramaniam Sankar
2:45 Trans-receiver with Digital Modulator & Frequency Up Converter for Secured Communication
Trans-receiver designed with programmable digital Modulators, Frequency up converter, RF Tuner & Digital demodulator can be used for secured communication. The proposed system will be useful for frequency Trans-modulation applications also for the frequency range of 50 MHz to 2.5 GHz for various signal types like DVBC, DVBT, ISTBT, NTSC etc. Secured communication between the systems are required in many applications which can be achieved using the proposed system. Simple Trans-modulation technique is used for the frequency conversion. Programmable Modulator & RF frequency up converter are used as Transmitter whereas Tuner &programmable Demodulator will function as receiver.The designed system is useful for any commercial or military applications.
3:15 BRAPTER: Compact Braille Transput Communicator
The invention is called Compact Braille Transput Communicator or BRAPTER in short. This invention provides visually impaired people the independence in writing and communicating in the form of Braille. The invention belongs to field of humanitarian and robotics. The invention consists of two parts. First part is the sensory input pad called as 'e-notebook' which takes the input in the form of Braille code, converts it into English and stores the same in SD card. The data stored in the SD card can be retrieved by a computer or a mobile. The second part is the embosser which prints the data stored in the SD card in Braille. The hardware equipped with the stylus punches out the required Braille code on an embossed paper. Thus this invention imprints the required Braille code. The first part is a device that fits on the palm and can be carried anywhere, thus helping them to take their notes at any part of time. This also gives them the liberty of having their own diaries in their pockets. This removes the burden of punching the notes manually which consumes lot of energy and time. The present invention help them to type their thoughts into the e-notebook. The embosser is designed to print the Braille code at high accuracy. It is available at very low cost compared to the existing technology, which can help people immensely, especially to the developing and underdeveloped countries, where major population belongs to middle or low class section. This can be handled by visually impaired people themselves. It is a user-friendly device unlike the big and complex embossers. It is also light weight unlike the heavy and stationed embossers. Portability makes it more versatile. The embosser is light and foldable to a size of a laptop ! Usage of the present product can establish a better teaching environment as teachers can effectively connect with the students. The product encourages visually impaired people to maintain their personal diary and have a normal lifestyle.

Saturday, October 7, 15:45 - 16:00

Tea and Network Breakgo to top

Room: Break and Lunch Area

Saturday, October 7, 16:00 - 16:45

Panel Discussion: "Innovation in the new age - Challenges, directions"go to top

Ball Room 1

Saturday, October 7, 16:45 - 17:15

Valedictory Function and Awards Ceremonygo to top

Ball Room 1