面向自标定室内定位基于遗传算法优化SVM的视距与非视距信号分类  

LOS and NLOS signal classification based on genetic algorithm optimized SVM and used for self⁃calibrating indoor localization

作　　者：阚瑞祥 王玫 KAN Ruixiang;WANG Mei(School of Information Science and Engineering,Guilin University of Technology,Guilin 541006,China)

机构地区：[1]桂林理工大学信息科学与工程学院,广西桂林541006

出　　处：《现代电子技术》2021年第1期49-52,共4页Modern Electronics Technique

基　　金：国家自然科学基金委员会(61771151);广西重点研发计划项目(桂科[AB17292058]);国家自然科学基金(61961010)。

摘　　要：对于室内定位,基于安卓手机平台,利用特定频段的声技术TPSN测距模型,以到达时间(TOA)作为重要参数,可以实现高效、高实时性定位。算法中涉及到视距(LOS)与非视距(NLOS)信号wav文件分类,之前定位系统中大都使用支持向量机(SVM)完成分类。兼顾安卓手机实际使用情况与实际需求,改进了特征提取算法,并且使用遗传算法(GA)从训练样本再选择、参数寻优以及每代种群的最优保存策略三方面对SVM进行升级,提升了SVM性能。For indoor positioning,the acoustic technology TPSN ranging model in the specific frequency band can be used to achieve high efficiency and real⁃time positioning on the basis of Android mobile phone platform,in which the time of arrival(TOA)is an important parameter.In the algorithm,the wav file classification of line of sight(LOS)signal and non line of sight(NLOS)signal will be involved.In most previous positioning systems,support vector machine(SVM)is used for the classification.In this paper,the feature extraction algorithm is improved while considering the actual use and demand of Android mobile phones,and the genetic algorithm(GA)is used to upgrade the performance of SVM in three aspects of training sample reselection,parameter optimization and optimal preservation strategy of each generation population.

关 键 词：声信号 室内定位 分类 支持向量机 遗传算法 最优保存策略 

分 类 号：TN911.1-34[电子电信—通信与信息系统] TP391[电子电信—信息与通信工程]