Comparative statistical analysis of indoor positioning using empirical data and indoor radio channel models

Indoor geolocation using maximum likelihood algorithms applied to the measured reference radio map of the received signal strength (RSS) in indoor areas has attracted tremendous attention in the recent published literature. In this paper we examine generation of reference radio maps using two different radio propagation models to replace on site measurements for calibration of these indoor geolocation systems. The first model used for radio propagation is a statistical model recommended by the IEEE 802.11 standardization committee and the second model is a simple building-specific two dimensional ray tracing algorithm. The empirical reference radio map of the RSS from seven access points in 257 locations is collected in the first floor of the Atwater Kent Laboratory, Worcester Polytechnic Institute which represents a typical indoor office area. Using the empirical radio map and the radio map generated from the results of radio propagation modeling, the performance of the nearest neighbor and maximum likelihood location estimation techniques are compared.