Joint Relay and Jammer Selection based on Deep Learning for Improving the Physical Layer Secrecy in Cooperative Networks

In this paper, we develop a deep learning model to enhance the physical layer security in cooperative networks with eavesdropper. Firstly, we apply the deep learning based scheme to classify and select one node as a relay to assist the transmitter and maximize the achievable secrecy rate. Then, we apply deep learning based scheme to select two nodes, joint relay and jammer selection. The first node is selected as a relay node for assisting the source to transmit its data with enhancing the security and reliability using a Decode-and-Forward protocol, the second node is chosen as a jammer node for creating intentional interference at the eavesdropper. The joint relay and jammer selection give us high secrecy rate due to the benefits form the jammer node compared to the relay selection without jamming. Compared to the conventional optimal selection schemes, optimal selection without jamming and with jamming, we show that the proposed deep learning based scheme can achieve the same secrecy performance with relatively small feedback overhead. Moreover, we apply the learning-based relay selection using Support Vector Machine (SVM) which giving a good performance with relatively low implementation complexity.