改进的蜂窝网室内定位匹配算法

针对无线信道的动态衰落特性,基于蜂窝网的室内定位存在较大误差,提出一种改进的蜂窝网室内定位匹配算法——基于主成分分析法(principal component analysis,PCA)的子空间匹配算法,不仅保证系统实时性,而且有效地剔除大误差点,提高定位精度.该算法利用无线蜂窝信号非视距传播造成的位置特性构建离线指纹数据库,根据在线接收信号从离线指纹库中提取子指纹库,利用PCA算法对在线实测数据及子指纹库进行有效地降维,构建子空间,并结合加权K近邻匹配算法(weighted K nearest neighborhood,WKNN)估计出多个位置坐标,利用3σ准则对这些位置做筛选,输出最终定位结果.实验结果表明,基于PCA的子空间匹配算法在保证定位实时性的前提下,能有效剔除大误差点,提高整体定位性能.

Improved indoor localization matching algorithm for cellular networks

Due to the dynamic fading characteristics of wireless channel,the localization performance will be dramatically deteriorated in cellular network.Therefore,a principal components analysis (PCA) based subspace matching algorithm for indoor localization within cellular networks is proposed,which not only guarantees the real-time localization requirement,but also effectively eliminates the large error points for the sake of accuracy enhancement.Specifically,the off-line fingerprint database is constructed firstly according to the positional characteristics caused by the non-line-of-sight propagation of the wireless cellular signal.Then the sub-fingerprint database is extracted in line with the received signal.Next,PCA algorithm is utilized to reduce the dimension of both on-line received signal strength and sub-fingerprint database,which then forms a new subspace.After that,the weighted K nearest neighbors (WKNN) is adopted for multiple-target localization.Finally,the targets are screened based on 3σ rule in order to determine the final location.Experimental results show that the proposed approach can effectively eliminate the outliers and thus improve the overall on-line positioning performance under the premise of ensuring the real-time positioning.