基于A3C的认知物联网通信干扰消除算法

A3C-Based Interference Cancellation Algorithm for Cognitive Internet of Things Communication

针对频谱资源干扰管理的智能化需求,提出一种基于异步优势行动者-评论家(A3C)的干扰消除算法,旨在应对认知物联网(CIoT)通信系统中由频谱资源共享引起的干扰问题。通过智能体的学习和优化,帮助次级用户(SU)在受到干扰影响时做出最优的决策,从而改善通信质量和系统性能。在该算法中,当SU遭受干扰影响通信质量时,智能体通过学习和优化,使SU能够根据当前的位置信息、发射功率、接收功率以及干扰程度选择最低干扰程度的行动,并执行该行动后获得的奖励。智能体通过尝试不同减少干扰的行动,并根据奖励的反馈调整策略,达到最大化定义干扰程度指标和信号质量指标的奖励函数的目的,从而最大程度地减少干扰对通信质量的影响。实验结果表明,与传统k-means算法以及深度递归Q网络(DRQN)和深度Q网络(DQN)优化算法相比,基于A3C的干扰消除算法具有更短的收敛时间、更高的执行效率以及更高的系统吞吐量,较3种基准方法在吞吐量性能上至少提高7%,能够有效地减少干扰对通信质量的不利影响。

To address the intelligent needs of spectrum resource interference management,in this study,an Asynchronous Advantage Actor-Critic(A3C)-based intelligent interference elimination algorithm is proposed to tackle the interference problem caused by spectrum resource sharing in a Cognitive Internet of Things(CIoT)communication system.Through learning and optimization of the agent,the algorithm helps Secondary Users(SU)make optimal decisions when affected by interference,thereby improving communication quality and system performance.When the communication quality of the SU is affected by interference,the learning and optimization of the agent enables the SU to choose the action with the lowest interference degree according to the current position information,transmit power,and receive power and interference degree,whereby a reward is offered after performing the action.Intelligent agents attempt different actions to reduce interference and adjust their strategies based on the reward feedback,thereby maximizing the reward function by considering the interference level and signal quality indicators and minimizing the impact of interference on communication quality.The simulation results show that compared with the traditional k-means,Deep Recurrent Q-Network(DRQN),and Deep Q-Network(DQN)optimization algorithms,the A3C-based interference cancellation algorithm has a shorter convergence time,higher execution efficiency,and at least 7%higher throughput performance than the three benchmark methods.This demonstrates that the proposed algorithm can effectively reduce the adverse effects of interference on communication quality.