Indoor location using received signal strength of IEEE 802.11b access point

Abstract

In this paper, the fingerprinting technique is employed to locate a mobile user inside a building. The fingerprint information, collected from real in-building measurements, is formed by three IEEE 802.11b access points' signal strength data received by the mobile user. Three different pattern-matching algorithms have been studied: the multi-layer perceptron (MLP) neural network, the generalized radial neural network (GRNN) and the K-nearest neighbours (KNN) algorithm. Their performances in terms of localization accuracy are compared on both training and testing data. Results show that the K-nearest neighbours gives the best localization accuracy. The effect of the measurement's grid spacing has also been investigated. Experimental results show that the localization accuracy increases when the grid spacing decreases. However, when the spacing reaches a certain threshold value, the accuracy starts to deteriorate. It can be shown that, in reality, the localization accuracy is improved even after the considered threshold value