Indoor Location Learning Over Wireless Fingerprinting System With Particle Markov Chain Model

Abstract

This paper describes research toward a tracking system for locating persons indoor based on low-cost Bluetooth Low Energy (BLE) beacons. Wireless fingerprinting based on BLE beacons has emerged as an increasingly popular solution for fine-grained indoor localization. Inspired by the idea of mobility tracking used in the cellular network, this paper proposes a BLE-based tracking system, designated as BTrack, to learn the location area (LA) of an indoor user based on the reported wireless fingerprinting combined with statistical analysis. We propose a new particle Markov chain model to evaluate the LA-level performance regarding the visibility area in an environment with large obstacles. In the presence of sight obstructions, BTrack is evaluated using a real-world test bed built in a library with tall bookshelves. The performance of the proposed system is evaluated in terms of the mean distance error and the LA prediction accuracy considering the direct line-of-sight. Compared with the existing methods, BTrack reduces the average localization error by 25% and improves the average prediction accuracy by more than 16% given a random mobility pattern.

Keywords

• People tracking
• resampling
• position accuracy
• Bluetooth
• beacons
• machine learning