Accuracy of Bluetooth based Indoor Positioning using different Pattern Recognition Techniques

Abstract

Object indoor location is a field that receives much research effort but that is lacking enough maturity for its integration in popular devices like mobile phones. This paper describes the results of an experiment carried out to compare different pattern recognition algorithms in order to process the information from a set of Bluetooth transmitters, located in fixed positions, with the aim of locating an object in a precise position. Our conclusion is that the best algorithms, among the five we tested, are random forests and model-based clustering, which gave accuracies around 90%. We have also conducted experiments to analyse the influence of the number of Bluetooth transmitters and to determine the sets of features with better performance. The proposed approach is simple and gives 90% of accuracy for locating objects with 1 m precision, making it suitable for a wide range of applications.