A multi-agent deep Q-learning-based joint relay and jammer selection in dual-hop wireless networks

Physical layer security (PLS) has cropped up as a promising solution to secure the wireless network. Cooperative communication is capable of improving the PLS, in addition to increasing the coverage area and reliability. It applies to diverse wireless systems, including long-term evolution (LTE) cellular systems, mobile ad hoc networks, and wireless sensor networks. The selection of relay and jammer nodes from the cluster of intermediate nodes can easily counter the strong eavesdroppers. Existing techniques of joint relay and jammer selection (JRJS) solve the optimization problem to find near-optimal secrecy. However, due to their computational complexity, most of these algorithms are not scalable for large networks. In this manuscript, we introduced the multi-agent deep Q-learning (MADQL) algorithm for secure joint relay and jammer selection in dual-hop wireless cooperative networks. The JRJS is transformed into a prediction-based problem and solved using deep Q-learning algorithms. The proposed reinforcement learning technique is model-free and best suited for situations where the exact global channel state information (CSI) for all the links is unavailable. The secrecy performance of the introduced algorithm is compared with the existing techniques. Simulation results confirmed that the MADQL-JRJS algorithm outperformed the existing algorithms.