基于K-means和KNCN算法的室内定位方法  

Research on Indoor Positioning Method Based on K-means and KNCN Algorithm

作　　者：戴玉帆 金宏平[1] Dai Yufan;Jin Hongping(Hubei University of Automotive Technology,Shiyan 442002,China)

机构地区：[1]湖北汽车工业学院,湖北十堰442002

出　　处：《湖北汽车工业学院学报》2024年第4期59-63,68,共6页Journal of Hubei University Of Automotive Technology

基　　金：国家自然科学基金(51475150)。

摘　　要：针对KNN算法中待定点的邻近参考点参数固定、定位不灵活且误差较大的问题,提出了基于K-means和KNCN算法的室内定位改进算法。通过对指纹库数据进行主特征RSSI的K-means聚类,根据阈值动态选取簇心相似度较高的多个簇类。利用主特征RSSI权重分配计算加权距离的KNCN算法,采用高斯加权距离与簇类数据进行匹配,确定未知位置。实验结果表明,改进算法与KNN和KNCN算法相比,平均定位精度分别提升了29.4%和3%;平均定位时间比KNCN算法缩短了约83.4%。In view of the problems of fixed parameters of adjacent reference points,inflexible position-ing,and large error in the K-nearest neighbor(KNN)algorithm,an improved indoor positioning algo-rithm based on K-means and K-nearest neighbor centroid(KNCN)was proposed.By performing K-means clustering on the received signal strength indicator(RSSI)of the main features in the fingerprint database,multiple clusters with high cluster center similarity were dynamically selected according to the threshold.The KNCN algorithm used the RSSI weight distribution of the main features to calculate the weighted distance and used the Gaussian weighted distance to match the cluster data,so as to deter-mine the unknown position.The experimental results show that the average positioning accuracy of the improved algorithm is improved by 29.4%and 3%compared with KNN and KNCN algorithms,and the average positioning time is shortened by about 83.4%compared with KNCN algorithm.

关 键 词：室内定位 K-近邻 K-均值聚类 高斯权重 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]