An Anti-Jamming Communication Scheme Based on Global Performance for Multi-Hop Cognitive Radio Networks

Cognitive Radio Networks (CRN) are vulnerable to jamming because of its complex architecture and protocol. Furthermore, smart jamming technology has developed greatly in recent years. Anti-jamming design in multi-hop CRN has become a crucial challenge. In this paper, we proposed a global performance based anti-jamming communication scheme. The scheme considers the uncertainty result from distributed architecture and radio environment. Firstly, route selection is taken into account in the anti-jamming scheme. Secondly, the dual-threshold responsibility rating is applied to the power adaptation. Thirdly, a single-agent reinforcement learning framework based on global performance is proposed, which adopts Dueling Deep Q-Network (DQN). Simulation results show that the anti-jamming performance of the proposed scheme is superior to traditional algorithm in both sweeping jamming mode and the smart jamming mode. Furthermore, the reliability and robustness of the network are significantly improved.