Exploiting Secure Multihop Transmission in Underlying Cognitive Radio Networks: Analysis and Deep Learning Approaches

This paper studies the impact of the eavesdropping attacks on secure multihop transmission in cognitive radio networks. When secondary users transmit confidential information, the eavesdroppers overhear this information collaboratively and independently, respectively. We derive the exact closed-form expression for secrecy outage probability (SOP) with different eavesdropping attacks. Additionally, we propose the deep neural network (DNN)-based framework to capture the relation between input and output. The numerical results show that a colluding attack can wiretap more information compared to a non-colluding attack. From the Monte-Carlo simulation, the closed-form expression and DNN-based framework are verified. The effects of the interference threshold, the number of hops, the number of eavesdroppers, and the secrecy target data rate are investigated, respectively.