5G移动通信系统性能分析与预测方法研究  

Research on Performance Analysis and Prediction Methods of 5G Mobile Communication System

作　　者：周新鹏 徐凌伟[1,2] ZHOU Xinpeng;XU Lingwei(School of Information Science&Technology,Qingdao University of Science&Technology,Qingdao 266061,China;Key Laboratory of Opto-technology and Intelligent Control,Ministry of Education,Lanzhou Jiaotong University,Lanzhou 730070,China)

机构地区：[1]青岛科技大学信息科学技术学院,山东青岛266061 [2]兰州交通大学光电技术与智能控制教育部重点实验室,甘肃兰州730070

出　　处：《聊城大学学报（自然科学版）》2021年第4期23-28,共6页Journal of Liaocheng University:Natural Science Edition

基　　金：国家自然科学基金项目(61901409);光电技术与智能控制教育部重点实验室(兰州交通大学)开放课题基金项目(KFKT2020-09);山东省自然科学基金项目(ZR2017BF023);山东省博士后创新项目(201703032);青岛科技大学引进人才科研启动基金(010029029)资助;。

摘　　要：由于复杂多变的通信环境,第五代(5G)移动通信面临着严峻挑战。为了实现5G移动通信质量的增强,分析和预测移动通信系统的平均符号误码率(average symbol error probability,ASEP)性能就非常重要。因此,分析与预测了移动通信系统ASEP性能。首先推导了ASEP的精确闭合表达式,对系统的ASEP性能进行了分析。然后基于广义回归(generalized regression,GR)神经网络,提出了一种ASEP性能智能预测方法。最后,和反径向基函数(radial basis function,RBF)神经网络,极限学习机(extreme learning machine,ELM),小波神经网络(wavelet neural network,WNN)等方法进行了比较,仿真结果表明:算法预测性能更好,预测精度提高了24.3%。Due to complex and variable environments,the fifth generation(5G)mobile communication faces serious challenges.In order to realize the enhancement of 5G mobile communication quality,it is important to analyze and predict the average symbol error probability(ASEP)performance of mobile communication system.Therefore,this paper studies the ASEP analysis and prediction methods.Firstly,the exact close expressions are derived,and ASEP performance is analyzed.Then,an generalized regression(GR)neural network-based ASEP performance prediction algorithm is proposed.Compared with the radial basis function(RBF)neural network,extreme learning machine(ELM),and wavelet neural network(WNN)methods,the Monte-Carlo results verify that our proposed method can obtain a better prediction effect,and the prediction accuracy increases 24.3%.

关 键 词：移动通信系统 ASEP性能 性能分析 性能预测 

分 类 号：TN929.5[电子电信—通信与信息系统]