基于核模糊C均值指纹库管理的WIFI室内定位方法

WIFI fingerprinting localization based on Kernel Fuzzy C-means Ⅱ Clustering

针对目前已有的基于指纹的WIFI室内定位指纹库的管理方法对野值和噪声的敏感性,提出一种基于核模糊C均值聚类的指纹库管理的室内定位方法.利用核函数将指纹库从低维空间映射到高维空间并结合模糊聚类方法在高维空间进行指纹库管理,并在管理后的指纹库上进行定位匹配.将指纹库映射到高维空间可以使指纹库中的数据线性可分,从而实现更好的聚类.核模糊C均值(KFCM-Ⅱ)的聚类鲁棒性能够降低聚类对噪声和野值的敏感性,从而保证系统的鲁棒性.在实测数据的实验中,将所提出的方法与基于K均值聚类和基于模糊C均值聚类的室内定位方法进行对比,实验结果表明,所提出的方法相较于K均值方法和模糊C均值方法聚类准确度分别提高了14.20%和10.58%,定位精度分别提高了26.98%和20.43%.

An indoor fingerprinting localization method based on kernel fuzzy C-means clustering was proposed to address sensitivity to noise and outliers of already existed methods of management on fingerprint database in indoor positioning system based on WIFI.The proposed method firstly mapped the fingerprint database from low-dimensional to high-dimensional space and was combined with fuzzy clustering in high-dimensional space to manage fingerprint database using kernel function,followed by positioning match to obtain user＇s location.With mapping fingerprint database to high-dimensional space,the data could be separated linearly in high-dimensional space to accomplish better clustering.In addition,KFCM-Ⅱ was robust and could lower the sensitivity results from noise and outliers to ensure robustness of the positioning system.In the experiments of processing measured data,the proposed method,management methods on fingerprint database based on K-means Clustering and Fuzzy C-means Clustering were compared,and the results show that the proposed method outperforms K-means method and Fuzzy Cmeans method by 14.20% and 10.58% on clustering accuracy respectively,leading to 26.98% and 20.43%improvements in positioning accuracy.