基于强化学习的D2D通信资源分配系统研究

Research on D2D Communication Resource Allocation System Based on Reinforcement Learning

D2D通信可以实现距离相近的两个设备,不需要基站中继而通过复用通信资源进行直接通信,这种方式尽管能够提高传输速率,但是用户之间存在的干扰也不能忽略。为解决这一问题,提出一种基于强化学习的D2D通信资源分配的系统,利用Q学习的方法,在动态环境下将动作-状态构建成Q值表,系统通过执行的动作产生回报值,不断更新Q值表,最终趋于收敛,获得最大化效益。系统通过Q值表选择使得利益最大化的动作执行,给蜂窝小区内的用户分配信道和功率等级,减少干扰。从仿真来看,在算法的控制下,资源利用率获得了大大提高,降低了干扰。

D2D communication enables two devices that are close to each other to communicate directly by multiplexing communication resources without a base station..Although it can increase the transmission rate,interference between users cannot be ignored.In order to solve this problem,this paper proposes a D2D communication resource allocation system based on reinforcement learning.Using Q learning,the action-state is constructed into a Q-value table in a dynamic environment.The system generates a reward value through the actions performed.The Q table is continuously updated,eventually converging,and maximizing benefits.The system selects the actions that maximize the benefits through the Q-value table,allocates channels and power levels to users in the cell,and reduces interference.From the simulation point of view,under the control of the algorithm,the resource utilization rate has been greatly improved,and the interference has been reduced.