无线通信中基于卷积神经网络的负载均衡系统  

Load Balancing System Based on Convolutional Neural Network in Wireless Communication

作　　者：朱明浩 李君 李正权 仲星[1] 张茜茜 沈国丽 ZHU Minghao;LI Jun;LI Zhengquan;ZHONG Xing;ZHANG Xixi;SHEN Guoli(School of Electronics and Information Engineering,Nanjing University of Information Science and Technology,Nanjing 210044;School of Electronic Information Engineering,Wuxi University,Wuxi 214105;Key Laboratory of Advanced Control of Light Industry Process,Ministry of Education,Jiangnan University,Wuxi 214122;State Key Laboratory of Network and Switching Technology,Beijing University of Posts and Telecommunications,Beijing 100876)

机构地区：[1]南京信息工程大学电子与信息工程学院,南京210044 [2]无锡学院电子信息工程学院,无锡214105 [3]江南大学轻工过程先进控制教育部重点实验室,无锡214122 [4]北京邮电大学网络与交换技术国家重点实验室,北京100876

出　　处：《计算机与数字工程》2023年第9期2007-2012,2194,共7页Computer & Digital Engineering

摘　　要：针对无线系统中基站负载不均衡问题,提出一种利用Python+Keras框架测试CNN模型的负载均衡方案,首先提出考虑基站的负载的优化问题,通过基于图论的库恩-马克(Kuhn-Munkres,KM)算法求解,并设计训练了CNN网络模型来学习信道状态信息到实现负载均衡的解之间的非线性映射关系。实验仿真表明,该网络模型可以自动提取信道状态信息的特征并激活链路,准确率达到98.6%,不仅取得了基站间更均衡的负载,计算复杂度也更低,实现了快速可靠的关联调度。Aiming at the problem of base station load imbalance in wireless systems,a user association scheme using Python+Keras framework to test CNN is proposed.First,the load of the base station is considered in the optimization problem,solved by the Kuhn-Munkres(KM)algorithm based on graph theory,and a user association algorithm based on CNN is designed and trained.Multiple sets of experimental simulations show that the network model can automatically extract the characteristics of the channel state information between the base station and the user and activate the link,with an accuracy rate of 98.6%.Not only a more balanced load between the base stations is achieved,but also computational complexity is lower,fast and reliable associated scheduling is realized.

关 键 词：无线通信 负载均衡 CNN 资源分配 用户关联 

分 类 号：TN929.5[电子电信—通信与信息系统]