多指纹融合和区域细化的WiFi室内定位方法

Multi-fingerprint fusion and area refinement for WiFi indoor positioning methods

针对复杂环境中单一指纹特征表示位置信息导致的定位鲁棒性差和精度低的问题,提出一种基于多指纹特征融合和区域细化的无线保真(WiFi)室内定位方法:对特征选择(ReliefF)算法进行改进,采用改进的ReliefF算法确定接收信号强度指示(RSSI)、信号变化率(Rate)、双曲位置指纹(HLF)和信号强度差(SSD)这4种单一指纹特征对位置信息的贡献,并对4种单一指纹特征进行加权融合,得到组合位置特征;然后提出一种基于组合位置特征数据变化率和k均值(k-means)算法的区域细化算法,在离线构建指纹数据库时对定位区域进行细化。实验结果表明,基于多指纹融合和区域细化的WiFi定位方法比采用单一特征的WiFi定位方法具有更高的定位精度、速度和鲁棒性。

Aiming at the problems of poor robustness and low accuracy in location determination caused by the representation of positional information with a single fingerprint feature in complex environments,the paper proposed a wireless fidelity(WiFi)indoor positioning method based on the fusion of multiple fingerprint features and area refinement:the feature selection(ReliefF)algorithm was improved to determine the contributions of four single fingerprint features,including received signal strength indicator(RSSI),signal change rate(Rate),hyperbolic location fingerprint(HLF)and signal strength difference(SSD),to positional information;and a weighted fusion of these four single fingerprint features was performed to obtain a combined positional feature;then,an area refinement algorithm based on the rate of change in combined positional feature data and the kmeans algorithm was put forward,which refines the positioning area during the offline construction of the fingerprint database.Experimental results showed that the WiFi positioning method based on multi-fingerprint fusion and area refinement could offer higher positioning accuracy,speed and robustness compared to the WiFi positioning methods using a single feature.