Anti-jamming radar waveform design for repeater jammer using reinforcement learning

The ever-evolving technology has given rise to a dynamic and intricate Radar environment, profoundly impacting warfare. Repeater jammers pose a significant threat to Radar systems by re-transmitting intercepted Radar signals with added noise or false information, thereby adding false targets in Radar detection process. Traditional Radar waveform designs often struggle to counter such jammers due to their adaptive and sophisticated nature. To address this challenge, this paper proposes utilizing a set of waveform during a coherent processing interval (CPI) which help cancel out any waveform received out of order. This requires a set of waveform which are orthogonal to each other in delay and Doppler. To design such a set of waveform, this paper proposes leveraging reinforcement learning (RL). The process involves offline training on simulated Radar environments, learning from the consequences of actions, and gradually improving its decision-making capabilities to generate effective anti-jamming waveform. This study presents experimental results demonstrating the effectiveness of the proposed approach in enhancing Radar system performance by mitigating the impact of Repeater jammers. Simulation results show that the proposed set of waveform have improved auto-correlation peak sidelobes as compared to Barker code and random sequences, while reducing cross-correlation sidelobes as compared to Gold codes and Golay codes.