A Cognitive Radar Anti-Jamming Strategy Generation Algorithm based on Dueling Double DQN

With the improvement of the cognitive ability of the jammer, the electromagnetic countermeasure environment has become more complex, posing significant challenges to the anti-jamming performance of cognitive radar. Cognitive radar’s anti-jamming capabilities often depends on the correct generation of anti-jamming strategies, which makes the strategy generation a research hotspot cognitive radar anti-jamming. This paper proposes a cognitive radar anti-jamming strategy generation algorithm based on Dueling Double DQN (D3QN). The algorithm utilizes the state-value function and advantage function to estimate the quality of each radar echo and signal, thereby reducing the update error of the policy network. Additionally, to improve the speed of strategy generation, the proposed algorithm directly inputs the original echo into the policy network. Simultaneously, a jammer model with different working modes and interference type is established, and the corresponding jamming strategy is generated by the received echo, which is closer to the real confrontation environment. The effectiveness and robustness of the proposed algorithm are verified through high-fidelity adversarial experiments with variable jamming strategies.