车联网中基于深度强化学习的高可靠资源分配算法

针对车联网环境下用户通信质量下降以及频谱资源紧张导致车辆与车辆（vehicle to vehicle，V2V）链路的关键信息传输难以满足高可靠性通信需求的问题，提出了一种基于深度强化学习（deep reinforcement learning，DRL）的高可靠资源分配算法。考虑干扰、传输时延和有效传输概率等约束条件，构建了车联网的可靠性保障优化问题；为了进一步保障V2V链路关键信息传输的可靠性，设计了压缩网络来压缩环境状态信息；根据可靠性保障优化问题设计了相应的奖励函数，并基于双深度Q网络（double deep Q-network，DDQN）设计了一种智能资源分配策略。仿真结果表明，所提算法能有效提高车联网的总速率，实现V2V链路关键信息的高可靠传输。

Deep reinforcement learning based high reliability resource allocation algorithm for internet of vehicles

The increasingly complex communication environment of the Internet of Vehicles and the increase in the number of vehicles bring about the quality reduction of user communication and the shortage of spectrum resources, which makes it difficult for the transmission of key information in the vehicle-to-vehicle (V2V) link to meet the requirements of high reliability communication. For the above problem, a high reliability resource allocation algorithm based on deep reinforcement learning (DRL) is proposed. Considering the constraints of interference, transmission delay, and effective transmission probability, the reliability guarantee optimization problem of the Internet of vehicles is constructed. To further guarantee the reliability of V2V link key information transmission, a compression network is designed based on deep learning to compress environment state information. The corresponding reward function is designed according to the reliability guarantee optimization problem, and an intelligent resource allocation strategy is proposed based on the double deep Q-network (DDQN). Simulation results show that the proposed algorithm effectively improves the total rate of the Internet of Vehicles and realizes the high reliability transmission of key information in the V2V link.