基于自适应修正曼哈顿距离的室内定位方法

Indoor Positioning Method Based on Adaptive Correction Manhattan Distance

关键词:  室内定位  WiFi指纹  相似度度量  信号波动  曼哈顿距离

Key words:  Indoor positioning  WiFi fingerprint  Similarity metric  Signal fluctuation  Manhattan distance

摘要:

在室内WiFi环境下，针对常见指纹匹配算法所忽略的信号波动问题，提出了一种基于自适应修正曼哈顿距离和AP选择的指纹匹配算法，并结合加权K近邻方法实现定位。首先采用AP选择算法获取部分受干扰程度小和出现频率高的AP，在指纹匹配时仅使用该部分AP的接收信号强度进行计算；在分析WiFi信号传播衰减公式和信号波动的基础上，提出了将自适应修正曼哈顿距离作为指纹匹配的度量距离，使用该距离旨在平滑信号波动对指纹相似度计算的影响；最后采用加权K近邻方法估计测试点的坐标。实验结果表明，在加权K近邻方法的框架下，基于自适应修正曼哈顿距离的定位算法在定位精度上优于基于欧氏距离、曼哈顿距离、余弦距离和Sorensen距离的定位算法。

Abstract:

In the indoor WiFi environment, a fingerprint matching algorithm based on the adaptive correction Manhattan distance (ACMD) and access point (AP) selection is proposed for the signal fluctuation problem neglected by the common fingerprint matching algorithms, and the weighted K nearest neighbor (WKNN) method is used to estimate the position. First, the AP selection algorithm is used to obtain reliable APs, and only received signal strengths (RSSs) from reliable APs are used for fingerprint matching. Second, after the WiFi signal propagation attenuation formula and signal fluctuation phenomenon are analyzed, the ACMD is proposed as a similarity metric, which is designed to smooth the effect of signal fluctuations on the calculation of fingerprint similarity. Finally, WKNN is used to estimate the coordinates of the Test Point. The experimental results show that in WKNN method, the proposed algorithm is better in positioning accuracy than other positioning algorithms using Euclidean distance, Manhattan distance, cosine distance or Sorensen distance.