Building information aided Wi-Fi fingerprinting positioning system

Abstract

In Wi-Fi fingerprinting localization systems, good positioning accuracy requires accurate fingerprint databases; however, the collection of accurate is a labor intensive work. Therefore, the collection, calibration and maintenance of fingerprints in Wi-Fi positioning have always been hot research topics. In this paper, a parameterized fingerprint optimization system is proposed to obtain an accurate fingerprint database. The pairs of RSS (received signal strength) and corresponding position are collected through path survey several times, which makes the collection process more efficient but brings more noise. The wall-induced wireless signal propagation theory taking floor plans into account is used to model the APs through the least squares. The predicted values from theory are combined with survey values for de-noising in fingerprint database. An experiment performed at indoor environment shows that compared with the original fingerprints, the positioning accuracy of the optimized fingerprints is improved by 0.6 m on average from many tests.