Robust wireless communication in adversarial settings

Abstract

Robust wireless communication is critical to the functioning of military, civilian and commercial systems. In this thesis, we explore the security of existing wireless communication systems in the presence of adversaries who are trying to interfere with ongoing communication. By broadcasting RF signals over an open medium, wireless networks make themselves vulnerable to jamming attacks. Furthermore, the growing demand for ubiquitous wireless technology has led to increasingly complex medium–access and resource–sharing mechanisms. This creates security bottlenecks for the whole system since an adversary can now focus all of its jamming on these control mechanisms in-turn enabling smart energy-efficient Denial of Service (DoS) attacks. In our work, we show that, with the help of readily available, low-cost, fully configurable software-defined radio platform, an adversary can deploy highly efficient DoS attacks by targeting these bottlenecks. We investigate their impact and devise energy-and-computationally efficient, resiliency mechanisms. In the past, robust communication relied on the existence of pre-shared secret information used by legitimate clients to access the control information sent by the server on specific communication channels. Here, we consider a scenario where the adversary knows this information. Adversaries can be compromised users or “traitors” who are trying to deny the control channel communication. This leads to very efficient attacks, since jamming control channel communication results in the disruption of the longterm communication. In response, we first present several adversary-resilient control–information sharing mechanisms that allow users to receive medium–access information in the presence of adversaries, without requiring any hardware modifications. Then, we propose a method - based on Spread Spectrum (SS) - which, with hardware support, allows longterm communication without the need to pre-share secrets. This method significantly degrades the effects and efficiency of smart jamming attacks. We analyze the efficiency of proposed solutions and evaluate their resilience under various jamming regimes, using a USRP/GNURadio/GPU-aided network platform. Finally, we study the resilience of the longterm IEEE802.11 communication against jamming attacks. We specifically look at attacks that exploit the vulnerabilities within the IEEE802.11 MAC protocols and rate-adaptation algorithms, and devise jamming-resilient mechanisms to address the identified weaknesses. We evaluate these mechanisms using analysis, simulation, and where applicable, real-world experiments with carefully designed system prototypes.