Radio map position inference algorithm for indoor positioning systems

Indoor positioning systems (IPS) have gain significant attention in the recent years; due to their relative low cost and high accuracy. However, till today, RSSI (received signal strength indicator)-based localization method pose a major challenge to engineers. The effects of severe fading and dynamic nature of the indoor environment greatly degrade the accuracy of the system. In this paper, a position inference algorithm using radio map is proposed to improve the accuracy of RSSI-based indoor locating systems. The radio map is first setup during the calibration phase; samples of RSSI at each point, within the area of interest, is recorded and converted into probability density function. During operation phase an inference algorithm, based on Bayesian probability and distance of the calibrated points involved, can determine the likely position of the object of interest that is between the calibrated points. The system yields an accuracy of less than 1.5 meter, which is better than the current RSSI-based localization system.