基于增强学习的D2D用户和蜂窝用户传输功率的联合优化  

Deep Reinforcement Learning⁃Based Joint Optimization for Transmitted Power of D2D User and Cellular User

作　　者：徐义晗 XU Yihan(School of Computer and Communication,Jiangsu Vocational College of Electronics and Information,Huaian Jiangsu 223003,China)

机构地区：[1]江苏电子信息职业学院计算机与通信学院,江苏淮安223003

出　　处：《电子器件》2024年第2期458-463,共6页Chinese Journal of Electron Devices

基　　金：淮安市创新服务能力建设计划项目-淮安市软件测试技术重点实验室(HAP201904)。

摘　　要：针对D2D用户与蜂窝用户间的干扰问题,提出基于深度增强学习的传输功率优化算法(DTPO)。DTPO算法通过调整传输设备的传输功率,缓解干扰。先将功率分配问题构建成基于线性约束的联合优化问题,再利用深度增强算法求解,获取D2D用户和蜂窝用户的传输功率,进而最大化和速率。仿真结果表明,DTPO算法的性能逼近于穷尽搜索算法的性能。Targeting at the interference between D2D user and cellular user in D2D communication underlay cellular network system,deep enhancement learning⁃based transmission power optimization(DTPO)algorithm is proposed.The interference is mitigated by opti⁃mizing the transmit power of the devices.The power allocation problem is generally modeled as a NP⁃hard combinatorial optimization problem with linear constraint.Then deep reinforcement learning(DRL)algorithm is used to optimize the transmit power for both D2D users and cellular users,and the sum⁃rate is maximized.Simulation results show that DTPO algorithm affords similar performance with exhaustive search algorithm.

关 键 词：支持D2D通信的蜂窝通信系统 干扰 传输功率 深度增强学习 和速率 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]