The Application of Manhattan Tangent Distance in Outdoor Fingerprint Localization

Received signal strength (RSS) is a typical measurement used for indoor and outdoor localization because of its low power consumption and no requirement of additional mode. Fingerprint localization is known to have higher accuracy than traditional methods and the most frequently used algorithm is K Nearest Neighbors (KNN), where common metrics have some disadvantages. In this paper, exponential transformation and tangent distance will be utilized to improve accuracy of weighted-KNN based fingerprint localization, where Manhattan tangent distance (MTD) and approximate Manhattan tangent distance (AMTD) are proposed. In addition, data are collected by a field tester from three different outdoor area for test. Experiments demonstrate that MTD achieves the lowest root mean square error (RMSE), AMTD achieves the second lowest RMSE with much lower computation complexity compared to MTD and both of them outperform common metrics.