Comparing Adversary Defense Mechanisms in Cognitive Radio Networks

Background: In a cognitive radio network, the cognitive transmitter senses the medium to detect spectrum opportunities and transmits its own data if the channel is sensed to be idle. A jammer can also sense the medium and identify the slots of successful transmission. The jammer’s main objective is to reduce the throughput of the cognitive transmitter. Methods: Towards this objective, the jammer builds a deep learning classifier in which the most recent sensing results of acknowledgments (ACKs) sent by the receiver are used to predict the slots of successful transmissions of the cognitive transmitter. This allows the attacker to reliably predict the successful transmissions and can effectively jam these transmissions. The deep learning classification soft decision probabilities are used by the jammer for power control subject to a certain power budget. A receiverbased defense mechanism is developed against jamming attacks. The receiver purposely takes some wrong actions, i.e., sends ACK when the transmission is not successful and vice versa, to poison the training process of the attacker. Results: We show that our receiver’s defense mechanism effectively enhances the throughput of the cognitive transmitter by about 25% when compared to the transmitter’s defense mechanism, where the transmitter takes some wrong decisions when it accesses the medium. Conclusion: A novel defense mechanism against jamming attacks in cognitive radio networks is introduced.