Modeling and Designing Network Defense against Control Channel Jamming Attacks: A Passivity-Based Approach

Abstract

Cyber-physical systems rely on distributed embedded wireless nodes for sensing, computation, and control, often leaving them vulnerable to intelligent attacks. Detecting and mitigating such attacks in a resource-efficient manner requires the network to identify the components under attack and adapt its response based on the observed network performance and attack impact. In this paper, we present our ongoing work on a passivity-based, control-theoretic framework for designing and evaluating adaptive network defenses against the control channel jamming attack. In this attack, an adversary makes use of information gathered from the compromised nodes to identify the location and disrupt the control messages. We present a dynamical model of the attack impact over time, and represent the network response as a feedback control action. Using passivity theory, we identify a class of linear controllers that correspond to network responses that guarantee convergence to a desired operating point. Through a numerical study, we analyze network characteristics such as robustness to uncertain observations of the network state, stability in the presence of delayed observations, and rate of convergence to the desired state. We find that, in general, higher order controllers provide increased robustness and stability, at the cost of slower overall convergence.