A fingerprint-based technique for indoor localization using fuzzy Least Squares Support Vector Machine

Abstract

Considering the growing demand of location-based services in indoor environments and development of Wi-Fi in recent years, indoor localization based on fingerprinting has attracted many researchers interest. In this paper, we introduce a novel fuzzy Least Squares Support Vector Machine (LS-SVM) based indoor fingerprinting system by using the received signal strength (RSS). In the offline phase, RSS values of all Wi-Fi signals detected from the available access points are collected at different reference points with known locations and are stored in a database. In the online phase, the target position is estimated by calculating fuzzy membership functions of samples and using formulation of fuzzy LS-SVM method. Simulation results show that average estimation error of the proposed method is 2.56m, while average positioning error of traditional LS-SVM methods was 4.61m.