An Adaptive Navigation Algorithm based on Reinforcement Learning for 5G Communication-Positioning Integrated Signal

Low power consumption receivers are an important research direction for positioning and navigation. This paper studies the low power consumption navigation technology of intermittent signal navigation of the positioning receiver of the 5G communication-positioning integrated signal (CPIS), and aiming at the filter divergence problem caused by the poor robustness of the noise covariance matrix of kalman filter algorithm in the intermittent signal navigation mode, the kalman filter based 5G CPIS navigation algorithm based on reinforcement learning is proposed, taking the kalman filter navigation algorithm as the environment, the negative direction of the position error as the reward, and the deep deterministic policy gradient (DDPG) is introduced. Obtain the optimal process noise covariance matrix estimate from the continuous operating space. Simulation shows that the algorithm effectively suppresses the filter divergence problem and improves the positioning robustness of the 5G pass-through fusion positioning receiver in the low power consumption navigation mode.