基于学习的能量采集认知M2M通信资源分配算法  

A Learning-Inspired Resource Allocation for Energy Harvesting-Powered Cognitive M2M Communications

作　　者：许艺瀚 田永波 张扬刚 花敏 周雯[1] XU Yi-han;TIAN Yong-bo;ZHANG Yang-gang;HUA Min;ZHOU Wen(College of Information Science and Technology,Nanjing Forestry University,Nanjing,Jiangsu 210037,China;School of Information Science and Technology,Fudan University,Shanghai 200433,China)

机构地区：[1]南京林业大学信息科学技术学院,江苏南京210037 [2]复旦大学信息科学与工程学院,上海200433

出　　处：《电子学报》2023年第2期467-476,共10页Acta Electronica Sinica

基　　金：国家自然科学基金(No.61801225,No.61601275);南京林业大学引进高层次人才和高层次留学回国人员科研基金(No.GXL015)。

摘　　要：本文针对能量采集认知机器到机器(Machine-to-Machine,M2M)通信的能量效率问题,在保证服务质量(Quality of Service,QoS)的条件下,提出了一种能效优化算法.以最大化网络中用户能效为目标,综合考虑传输功率控制、时隙分配、传输模式选择、中继选择以及每个设备的能量状态为约束,将优化问题建模为一个混合整数非线性规划问题.将该能效优化问题转化为离散时间有限状态马尔科夫决策过程(Discrete-time and Finite-state Markov Decision Process,DFMDP)进行求解.提出一种基于深度强化学习的算法寻找最优策略.仿真结果表明,所提算法在平均能效方面优于其他方案,且收敛速度在可接受范围内.In order to optimize the energy efficiency for energy harvesting-powered cognitive M2M communications underlaying cellular network,an energy efficient algorithm is proposed while guaranteeing the quality of service of users.Firstly,the problem is formulated as a mixed integer nonlinear programming problem with the goal of maximizing energy efficiency by jointly considering transmission power control,time slot allocation,transmission mode and relay selection with the constraints of the energy status of each device.After that,the optimization problem is modeled as a discrete-time and finite-state Markov decision process.Afterward,a deep reinforcement learning-based algorithm is proposed to find the optimal strategy.Numerical results validate that the proposed scheme outperforms other schemes in terms of average energy efficiency with an acceptable convergence speed.

关 键 词：能量收集 认知无线电 M2M通信 资源分配 深度强化学习 

分 类 号：TN914[电子电信—通信与信息系统]