Improved fingerprint algorithm for WLAN-based vehicle positioning

Reliable and accurate vehicle position information is important for autonomous systems. The increased popularity of wireless networks has enabled the development of positioning techniques that rely on WLAN signal strength. Fingerprint architecture is one of the most viable solutions for Received Signal Strength (RSS)-based positioning. The most challenging aspect of the fingerprint based method is to formulate a distance calculation that can measure similarity between the observed RSS and the known RSS fingerprints. In this paper we proposed an improved fingerprint-based algorithm that incorporates the spatial diversity and the road constraints for vehicle positioning. We present how to apply the properties of the spatial diversity and the road constraints to enhance the robustness and the accuracy of the fingerprint algorithm. To further improve the accuracy of the fingerprint algorithm, a Dead Reckoning (DR) system, which has been widely used for vehicle navigation, has been integrated with the WLAN-based positioning system. We have conducted extensive field tests and simulations for the proposed positioning algorithm, and key outcomes are given out. It has been demonstrated by the results that our algorithm could significantly improve the performance of WLAN-based vehicle positioning system.