Adaptive Modulation and Coding based on Data-Driven Learning Algorithm in Polar HF OFDM-CPM System

Adaptive modulation coding (AMC) technology can automatically select the most effective modulation scheme and coding rate according to channel conditions, so as to improve the power utilization and spectrum utilization of the system. Aiming at the problem of low and inefficient traditional Q-learning due to the large number of continuous channel states in polar HF communication processes. In this paper, a deep reinforcement learning adaptive coding and modulation scheme based on channel classification is proposed in OFDM-CPM system. Firstly, the correlation analysis of the received pilot signal is carried out to roughly classify the measured or simulated channels. At this time, the channel state is reduced to a less discrete state by the feature. Finally, the Double Deep Q-network(DDQN) is used to complete the adaptive process of the modulation and coding parameters at the transmitter. The simulation results show that the proposed method has faster convergence speed, higher cumulative reward and higher spectral efficiency than the DDQN learning scheme.