基于深度强化学习的多RIS辅助干扰信道传输方法  

Transmit Design for Multi-RIS-Assisted Inference Channel:A DRL Approach

作　　者：侯懿圃 郭为秀 李全鹏 张胜利 陈帝 陆杨 HOU Yipu;GUO Weixiu;LI Quanpeng;ZHANG Shengli;CHEN Di;LU Yang(School of Computer Science and Technology,Beijing Jiaotong University,Beijing 100044,China)

机构地区：[1]北京交通大学计算机科学与技术学院,北京100044

出　　处：《移动通信》2024年第8期85-89,101,共6页Mobile Communications

基　　金：国家自然科学基金“混合能量源驱动的‘去蜂窝’网络设计理论与方法”(62101025);北京市科技新星计划“基于‘去蜂窝’网络的无线覆盖扩展技术研究”(Z211100002121139);北京市自然科学基金-丰台轨道交通前沿研究联合基金“面向跨线运营的轨道交通站-车-网通信异制协同控制方法”(L221010)。

摘　　要：无线覆盖扩展是6G的核心任务。RIS作为新兴智能无线环境技术,可以有效缓解障碍物对高频信号传播的影响。然而,RIS通信系统也面临信道估计困难和资源优化复杂的挑战。为此,研究面向多RIS辅助干扰信道系统的能量效率最大化问题。为降低信道估计频次,考虑信道统计信息;为求解所提出问题,采用深度强化学习方法。将所考虑问题建模为MDP,并提出混合最大合并比和破零方法设计动作空间;在此基础上,基于近端策略优化算法训练智能体求解MDP。实验结果展示了所提出方法在不同信噪比下的收敛情况与适应能力。Enhancing wireless coverage is the core task of the sixth-generation mobile communication system(6G).Reconfigurable intelligent surface(RIS),as an emerging smart radio environment enabler,is capability of effectively alleviating the impact of obstacles on the high-frequency signal propagation.However,RIS communication systems also face challenges in channel estimation and complex resource optimization.Therefore,this paper studies the energy efficiency maximization problem for multi-RIS-assisted interference channel systems.To avoid frequent channel estimation,the channel statistical information for RIS-user links is considered.To solve the proposed problem,a deep reinforcement learning method is adopted.The problem is modeled as a Markov decision process(MDP)and a hybrid method of the maximum ratio combining and zero forcing is proposed to design the action space.Then,the proximal policy optimization algorithm is used to train the agent to solve the MDP.Experimental results demonstrate the convergence and adaptability of the proposed method under different signal-to-noise ratios.

关 键 词：智能超表面 干扰信道 能量效率 深度强化学习 

分 类 号：TN929.5[电子电信—通信与信息系统]