Reinforcement Learning Based Agile Transmission Strategy for Netted Radar Anti-Jamming

Anti-jamming is an urgent yet challenging task in the rapidly changing world of electronic warefare. In an attempt to avoid jamming, frequency agile radars based on reinforcement learning are used to learn the environment and generate countermeasures. Current reinforcement learning based anti-jamming methods are constrained to single radar anti-jamming or assume a weak jamming environment. To evade these limits, we propose a reinforcement learning based agile transmission strategy for netted radar anti-jamming. Through the help of reinforcement learning, the radar network can learn the jammer's strategy and coordinate their actions to reach the optimal detection performance. In the proposed method, we devise a fusion model based on the agile transmission strategy. Also, a reward function is established with reference to the detection probability. Then, the radar network anti-jamming decision process is modeled as a Markov decision process (MDP). As a typical algorithm to solve discrete problem in reinforcement learning, Q-learning is employed to learn the optimal strategy in the electronic countermeasure environment. Simulation results are presented to demonstrate the effectiveness of the proposed method.