Optimization of Wireless Resource Management Algorithm Based on Deep Q-Network

Abstract

With the continuous and rapid growth of the number of access devices, traditional resource management algorithms can no longer meet the actual needs. In order to solve the problems of poor coordination of wireless network resources and uncoordinated service quality in traditional wireless resource management, a new wireless network resource management algorithm is proposed based on the actual network structure of the Industrial Internet of Things (IIoT), combined with the Markov decision-making process and the improved deep Q network algorithm. The results show that the performance of the proposed algorithm is significantly improved after training, surpassing the time limit of Aloha algorithm and random algorithm, achieving high channel utilization and system rate, and the convergence speed is faster than that of traditional deep Q network algorithms. When the arrival rate is greater than 1.0, the channel rate and average sum rate of the research algorithm reach 94% and 25bits/s/Hz, respectively, which are significantly better than the comparison algorithms. As the bandwidth of the system increases, the latency of each algorithm decreases. Among them, the performance of the research algorithm is better than that of the comparison algorithm, the system latency and computational cost are lower, and the system latency is the lowest under different numbers of blockchain nodes. The results show that the proposed resource management algorithm can realize IIoT dynamic wireless resource management and promote the efficient utilization of wireless network resources.