Abstract
Current indoor location determination systems require an offline training phase in order to build a radio map that contains a list of radio fingerprints at known locations. The process of building this radio map is usually a time-consuming and requires an intensive work where the accuracy of these systems depends on various factors such as the number of collected radio fingerprints and the time in which they were collected. This paper presents an indoor location determination system that does not require a time-consuming offline training phase in order to build the radio map. Instead, the system uses the online phase for both reading the current Received Signal Strength (RSS) and inferring the location values for each random variable using Bayesian Graphical Model (BGM). The proposed system is based on Markov Chain Monte Carlo (MCMC) sampling techniques in order to draw samples from the posterior distribution. We will also introduce the Feed and Infer algorithm to be used in conjunction with the proposed graphical model. The proposed system will be compared with other single-phase systems.