Resource Allocation for V2X Assisted Automotive Radar System based on Reinforcement Learning

Due to the complexity and variability of road traffic scenarios and the existence of blind spots in car radar detection, radar performance of single vehicle is limited. To address this issue, we propose a joint communication and radar sensing (JCR) system for Intelligent Connected Vehicles (ICVs), where communication is used to assist in reducing the miss detection probability. Based on this, we model the time resource allocation problem as a Markov Decision Process (MDP), and design the Q-learning and the Double Deep Q-learning Network (DDQN) algorithms to optimize the allocation of time resources for radar and communication functions dynamically. The simulation results show that compared with the Round-robin algorithm, the Q-learning and the DDQN algorithms can increase the communication data throughput by more than 6.6% and reduce the miss detection probability by more than 29.4%. The miss detection probability of the system using the assisted mode is 10.7%-17.2% lower than that of the system without it.