Indoor Localization in Wireless Networks Based on a Two-Modes Gaussian Mixture Model

This paper presents and evaluates a method to localize devices that communicate using a wireless network. The distances that separate a blind node, willing to determine its position, and a set of anchor nodes, that know their locations, are evaluated using the signal attenuation (RSSI) measured on data packets. However, multipath effects, frequent in an indoor scenario, introduce randomness in signal propagation, reducing localization accuracy. We propose to use a maximum likelihood estimator on a two-modes Gaussian Mixture model approach to detect and exclude outlier measurements. We evaluate and compare this method using experimental measurements.