Research On Wireless Resource Allocation Based On Reinforcement Learning For High Speed Train

With the increase in the number of high-speed rail passengers,the requirements for information transmission speed and wireless network service level have gradually increased.In order to cope with the problems of high-speed mobile communication systems,especially for the problem of excessively high system energy consumption,more and more 5G communication technologies are being used in high-speed scenarios.In order to more efficiently allocate wireless resources in high-speed railway communication scenarios,this paper studies the resource allocation of high-speed railway wireless communication systems and proposes resource allocation algorithms for such scenarios.The key tasks in this article include the following:(1)The problem of resource allocation based on a single mobile relay system in high-speed rail scenarios is studied.First,we analyze the channel characteristics of the high-speed wireless communication system,consider the problem of co-frequency interference in the high-speed railway network with mobile relays,and find the expression of each user throughput of the system.Traditional convex optimization is hard to solve non-convex problem,so the Q-Learning algorithm is used to solve the problem of power allocation to ensure that the minimum transmission speed requirements of ground users can be met under the premise of lower complexity The power distribution strategy that takes into account the fairness between train users and ground users has achieved the goal of maximizing the throughput of train users.(2)This paper studied the problem of energy efficiency resource allocation in a multi-relay multi-user system under high-speed rail.First,we take channel conditions of the two-hop link of the high-speed rail into account,the throughput expression of the train user and the entire system is obtained.In order to meeting the minimum communication demands of train users,this problem is modeled as a nonlinear programming problem with maximum energy efficiency.Reinforcement learning algorithm is used to reduce the complexity.At the same time,through the established Q table,the best power distribution scheme can be obtained in real time.Simulation experiments show that in the high-speed rail scenario,the algorithm in this paper performs well in throughput performance and energy efficiency performance.