Dogfight in Spectrum: Jamming and Anti-Jamming in Multichannel Cognitive Radio Systems

Primary user emulation attack in multichannel cognitive radio systems is discussed. An attacker is assumed to be able to send primary-user-like signals during spectrum sensing period and thus jam the secondary user sensing the corresponding channels. An anti-jamming approach for the secondary user is to do random hopping over the multiple channels. When different channels have different qualities (e.g. probabilities of being idle), the secondary user needs to find the optimal tradeoff between choosing good channels and avoiding the jamming of the attacker. The procedure of jamming and anti-jamming is coined dogfight in spectrum due to its nature of pursuit and evasion. For the one-stage case, the dogfight is modeled as a normal zero-sum game and the Nash equilibrium strategy (max-min point) is obtained. For the case of multi-stage game, the dogfight is essentially a stochastic game with partial observations and imperfect monitoring. The game is analyzed by fixing the strategy of the secondary user and finding the optimal attacking strategy of the attacker using the framework of partially observable Markov decision process (POMDP). The optimal strategies are numerically computed for both one-stage and multi-stage games.