Performance analysis of deep reinforcement learning-based intelligent cooperative jamming method confronting multi-functional networked radar

Abstract

With the development of artificial intelligence technology, more and more intelligent countermeasure methods are applied in military confrontation fields to improve the intelligent level of weapons. Traditional radar jammers generate different jamming types by template matching, game theory or reasoning, which lack intelligent and adaptive jamming strategies in the battlefield environment with intelligent confrontation. To solve the intelligent decision-making problem of jammers in radar countermeasure, a cooperative jamming decision-making (CJDM) method based on reinforcement learning (RL) is proposed in this paper. The double deep Q network based on priority experience replay (PER-DDQN) is brought into the cooperative jamming strategy, and the CJDM model based on PER-DDQN is established in this paper. The scene of multiple jammers against multi-functional networked radar was built to simulate and analyze the performance of the proposed CJDM model based on PER-DDQN. The simulation results show that the proposed PER-DDQN can overcome the problem of data correlation and avoid unnecessary iteration, which is more suitable for sparse reward environment compared with deep Q network (DQN). Meanwhile, the proposed CJDM method based on PER-DDQN can effectively and intelligently realize optimal jamming decision-making.