Trust based channel preference in cognitive radio networks under collaborative selfish attacks

Secondary spectrum data falsification (SSDF) is a common attack in cognitive radio networks, where dishonest nodes share spurious local sensing data. This behavior misleads the collective inference on spectrum occupancy. The situation is more aggravated when a collaborative SSDF attack is launched by a coalition of selfish nodes. Defense against such collaborative attacks is difficult with popularly used voting based inference models. This paper proposes a method based on Bayesian inference that indicates how much the collective decision on a channel's occupancy can be trusted. Using an anomaly monitoring technique, we check if the reports sent by a node match with the expected occupancy and classify the outcomes into three categories: i) if there is a match, ii) if there is a mismatch, and iii) if it cannot be decided. Based on the measured observations over time, we estimate the parameters of the hypothesis of match and mismatch events using a multinomial Bayesian based inference. We quantitatively define the trust as the difference between the posterior beliefs associated with matches and that of mismatches. The posterior beliefs are updated based on a weighted average of the prior information on the belief itself and the recently observed data. We conduct simulation experiments that show that the proposed trust model is able to distinguish the attacked channels from the non-attacked ones. Also, a node is able to rank the channels based on how trustworthy the inference on a channel is. We are also able to show that attacked channels have significantly lower trust values than channels that are not.