Workshop on Cooperative and Cognitive Networks (CoCoNet7)

Wireless multi-hop networks are vulnerable to various attacks including, most notoriously, denial-of-service (DoS) attacks. Being exposed on the propagation channel, wireless transmissions can easily be degraded by malicious jammers. In traditional wireless security, jamming attacks are typically dealt with in isolation and independently of other attacks. While this divide-and-conquer approach has contributed to facilitating the analysis and design of effective countermeasures, it fails to capture the new challenges brought by heterogeneous cross-layer attacks. By utilizing multiple means at multiple layers of the protocol stack with a unified goal, the adversary may be able to obtain an unexpected edge. Traditional security approaches, however, fail to address the potential consequences of such heterogeneous cross-layer attacks.In this talk, we will provide an overview of recent work aimed at developing a cooperative framework for assured high-throughput networking with heterogeneous cross-layer attacks. Our objective is to develop a unified universal framework capturing potential attacks at all layers to address the corresponding vulnerabilities.We will first discuss a new mathematical framework to develop optimal cooperative anti-jamming strategies. While traditional anti-jamming schemes only exploit frequency diversity, we introduce the notion of cooperative anti-jamming, where denial-of-service attacks are dealt with by leveraging cooperation among different legitimate users. Two levels of cooperation are considered. At the physical layer, cooperative communication based on relaying is used to enhance the link throughput of jammed links; at the MAC layer, legitimate users cooperate to optimally regulate their channel access probabilities so that jammed users gain a higher share of channel utilization. The problem of developing optimal cooperation strategies is formulated as a distributed pricing-based optimization problem and a best response algorithm with provable convergence is proposed. We show that the proposed framework achieves close-to-global optimality.We will then discuss how to integrate network layer and data security requirements into the framework. Most routing protocols for wireless multi-hop networks are based on controlling the network dynamics at multiple layers. This coupling implies that changes in the dynamics at layers other than the network layer also affect the routing decisions. As a result, an adversary may be able to extend attacks to the network layer by impacting functionalities at the physical layer through jamming. A jammer can in fact degrade the capacity of selected communication links. Then, because cross-layer routing protocols are often designed to avoid links with low capacity, jammed nodes are likely to redirect their traffic by choosing alternate paths. With proper collaboration from compromised nodes, data may directly to compromised nodes. We model hammer and attack and analyze the threat to wireless multi-hop networks. Characteristics helpful in countermeasure designs are observed, and, based on such observations, we design a trust-based cooperative countermeasure that shows considerable resistance to the attack.