User-centric view of smart attacks in wireless networks

Abstract

By applying smart and programmable radio devices, selfish end-users can launch smart attacks and choose multiple types of attacks such as jamming and eavesdropping according to the ongoing transmission of wireless networks. In this paper, we apply prospect theory (PT) to formulate the interaction between a smart attacker as an end-user who makes subjective decision regarding his or her attack mode under uncertain attack detection accuracy and a mobile user who chooses whether or not to apply the higher-layer security mechanism to enhance the physical (PHY)-layer security mechanism as a zero-sum game. The Nash equilibria (NEs) of the static smart attack game are derived and their existence conditions are provided to study the impact of the subjectivity of smart attackers. We also propose a defense strategy based on Q-learning, a model free reinforcement learning technique against subjective smart attacks in the dynamic game. Simulation results show that the proposed defense strategy can exploit the subjective view of smart attackers to suppress the attack motivation of end-users and improve the utility of the mobile user.