Adaptive Waveform Selection Algorithm based on Reinforcement Learning for Cognitive Radar

Cognitive radar is a newly emerging intelligent radar that can adaptively change the transmitted signal waveform according to changes in the target and environment to improve the accuracy of target state estimation. In this paper, the running process of cognitive radar adaptive transmission is analyzed, the tracking waveform parameter selection is correlated with the target state estimation and the reinforcement learning model is established. The problem of unknown target state space is solved by the “prioritized sweeping” method and the computational efficiency is improved by replacing “eligibility trace”. The simulation results show that the indirect reinforcement learning method is better than the fixed waveform and the waveform selection algorithm based on the minimum mean square error for the tracking accuracy and state estimation error of the target.