基于融合聚类的WiFi指纹定位算法研究
首发时间:2021-04-07
摘要:针对无线定位技术部署的成本因素、难易程度以及RSS室内指纹定位中相似指纹易误匹配的问题,采用基于融合聚类的WiFi位置指纹定位算法。该方法利用模糊C均值(Fuzzy C-Means,FCM)聚类代替了传统的硬聚类算法,不仅可以合理估算聚类中心RSS特征,避免硬聚类算法产生误差,还增加参考点之间的差异性,同时还能够降低特征匹配时的复杂度;针对FCM算法易受初始值影响易出现收敛于局部极值的问题,由于密度峰值聚类(density peaks clustering,DPC)算法能够准确表征聚类初始中心,故利用该算法能够有效弥补FCM算法的不足,从而可以提高定位精度。实验测试结果表明,基于融合FCM和DPC的定位算法比基于传统聚类的定位算法精度更高。
For information in English, please click here
Research on WiFi Fingerprint Location Algorithm Based on Fusion Clustering
Abstract:Wireless location technology is expensive and difficult to deploy, and similar fingerprints in RSS indoor fingerprint location are prone to mismatching. Aiming at the problem, the WiFi fingerprint locating algorithm based on fusion clustering was proposed. Fuzzy C-means (FCM) clustering is adopted instead of the traditional hard clustering algorithm. Because FCM algorithm can reasonably estimate the characteristics of RSS in clustering center and avoid errors generated by hard clustering algorithm. It also can increase the difference between the reference points and reduce the complexity of feature matching. But FCM algorithm tends to converge to local extreme values under the influence of initial values. Aiming at the problem, this paper use density peaks clustering (DPC) algorithm to effectively correct the shortcomings of FCM algorithm. Because DPC algorithm can accurately characterize the initial center of clustering. Therefore this way can improve the positioning accuracy. The experimental results show that the localization algorithm based on FCM and DPC is more accurate than that based on traditional clustering.
Keywords: Indoor location Fuzzy C-means clustering density peaks clustering algorithm
引用
No.****
动态公开评议
共计0人参与
勘误表
基于融合聚类的WiFi指纹定位算法研究
评论
全部评论0/1000