Analysis and Simulation of Sensing Deception in Fading Cognitive Radio Networks

We are interested in determining the sensitivity of a tactical cognitive radio (CR) system to intentional spoofing. That is, we assume the existence of an intelligent adversary whose goal is to attack the CR system by deceiving the secondary users into believing that as many frequency bands as possible are occupied by primary users, thus minimizing the number of bands in which the secondary users attempt to transmit. We refer to this operation by the adversary as "spoofing", and the specific spoofing signal we choose is a partial-band noise waveform. That is, for a given total power level that is available to the adversary, we maximize the average number of false detections incurred by secondary users as a result of the spoofing. We consider a channel such that each band experiences flat Rayleigh fading, whereby the fading is independent from band to band, and derive the average number of false detections by the secondary users due to the spoofing. The results obtained for the fading channel are compared to similar results for an additive white Gaussian noise channel (AWGN). They are also compared to a physically unrealizable scenario whereby the spoofing knows the instantaneous fade gains of the spoofing waveform at the victim CR receiver. This latter result is presented as a "worst-case" perspective as to how well the spoofing operation can be expected to perform.