Abstract:

The adaptive waveform selector is an important part of intelligent transmitters in cognitive radar.  Effective  waveform selection can transmit an optimal waveform sequence in different environments so as to track targets with higher accuracy. The problem of adaptive waveform selection is modeled as a stochastic dynamic model, and a Q-learning method is proposed to solve this problem under the fact that the transition probabilities of radar targets are unknown. The simulation results demonstrate that the proposed algorithm approaches the optimal waveform selection scheme and has a lower uncertainty of state estimation compared with the fixed waveform.