Smart probabilistic approach with RSSI fingerprinting for indoor localization

Abstract

This paper introduces an efficient probabilistic approach with RSSI fingerprinting for Indoor Localization. A Shannon's Entropy based access points (APs) selection is considered. Once the APs selection is performed, a probability is assigned to each training fingerprint based on RSSI measurements. Then, the user's location is estimated as a combination of training positions weighted with their corresponding probabilities. The proposed approach is performed on the UJIndoorLoc database. It shows good performances with lower computing complexity compared to others studied in literature.