| 摘要: |
| 在无蜂窝大规模多输入多输出（Multiple-Input Multiple-Output，MIMO）系统中，所有接入点（Access Point，AP）为用户提供服务的方式会造成较大的功率消耗和增加前传链路压力。以“用户为中心”的AP选择方案为每个用户选择最佳的AP集合,使系统频谱效率和能量效率得以提升。为了给每个用户选择最佳的AP集合，提出了一种基于反向传播（Back Propagation，BP）神经网络的接入点动态选择算法。将AP与用户之间的大尺度衰落向量作为输入，AP与用户之间的连接关系作为BP神经网络输出的预测向量，同时，针对BP神经网络初始化权值，提出了一种改进的自适应遗传算法以提高算法的收敛速度和收敛性能。仿真结果表明，改进后的BP神经网络预测结果更接近于实际值，并且比全连接和启发式的AP选择方案具有更高的系统和速率。 |
| 关键词: 无蜂窝大规模MIMO 接入点选择 遗传算法 BP神经网络 |
| DOI：10.20079/j.issn.1001-893x.221129002 |
|
| 基金项目:国家科技重大专项（2018ZX03001026-002） |
|
| A Dynamic Access Point Selection Algorithm in Cell-free Massive MIMO Systems |
| SHEN Min,QIU Deshi |
| (School of Communication and Information Engineering,Chongqing University of Posts and Telecommunications,Chongqing 400065,China) |
| Abstract: |
| In a cell-free massive multiple-input multiple-output（MIMO） system,all access points(APs) provide services to users,resulting in high power consumption and increased pressure on the fronthaul link.The “user-centric” AP selection scheme selects the best AP set for each user,which improves the system spectrum efficiency and energy efficiency.In order to select the best AP set for each user,a dynamic access point selection algorithm based on Back Propagation(BP) neural network is proposed.The large-scale fading vector between the AP and the user is used as the input,and the connection relationship between the AP and the user is taken as the prediction vector of the output of BP neural network.At the same time,an improved adaptive genetic algorithm is proposed for the initial weight of BP neural network to improve the convergence speed and convergence performance.Simulations show that the predicted result of the improved BP neural network is closer to the actual value,and it has higher system sum rate than fully connected and heuristic AP selection schemes. |
| Key words: cell-free massive MIMO access point selection genetic algorithm BP neural network |
