D2D异构网络节能模式选择和资源分配  

D2D heterogeneous network energy saving mode selection and resource allocation

作　　者：张世良 邹文景 唐良运 ZHANG Shi-liang;ZOU Wen-jing;TANG Liang-yun(Platform Security Branch,China Southern Power Grid Digital Grid Research Institute Limited Company,Guangzhou 510700,China)

机构地区：[1]南方电网数字电网研究院有限公司平台安全分公司,广东广州510700

出　　处：《计算机工程与设计》2023年第3期677-684,共8页Computer Engineering and Design

基　　金：广东省科技厅基金项目(GH13201906)。

摘　　要：为合理分配蜂窝网络的频谱资源,提升蜂窝网络能源效率,降低用户间的干扰,提出一种基于深度强化学习的设备到设备(D2D)异构网络节能模式选择和资源分配方法。构建系统模型并对节能模式选择和资源分配进行优化;将优化问题转化为马尔可夫决策过程(MDP),采用深度确定性策略梯度算法(DDPG)找到最优问题的最优策略,实现最大化长期能效。通过仿真分析与其它4种方法进行性能对比,实验结果表明,所提方法在D2D异构网络中具有更高的能源效率,表现出更好的收敛性,可有效提升系统吞吐量和频谱资源利用率。To reasonably allocate the spectrum resources of the cellular network,improve the energy efficiency of the cellular network,and reduce the interference between users,a device-to-device(D2D)heterogeneous network energy-saving mode selection and resource allocation method based on deep reinforcement learning was proposed.The system model was constructed and the energy-saving mode selection and resource allocation were optimized.The optimization problem was transformed into Markov decision process(MDP),and the deep deterministic strategy gradient algorithm(DDPG)was used to find the optimal strategy of the optimal problem to achieve maximum long-term energy efficiency.The performance was compared with the other four methods through simulation analysis.Experimental results show that the proposed method has higher energy efficiency in D2D heterogeneous networks with better convergence,which can effectively improve system throughput and spectrum resource utilization.

关 键 词：能源效率 异构网络 设备到设备 模式选择 资源分配 节能模式 深度强化学习 

分 类 号：TN929.5[电子电信—通信与信息系统]