Room-Level Wi-Fi Location Tracking

Context-aware applications for indoor intelligent environments require an appropriately accurate and stable interior positioning system to adapt services to the location of a mobile user or mobile device in a building. Different technologies provide a varying mix of resolution, accuracy, stability and challenges. In this paper we report on our experience using an existing Wi-Fi infrastructure without specialized hardware added to support location tracking. There are several approaches to track the location of Wi-Fi enabled devices within a building such as signal propagation models and signature matching. We found signature matching most effective in our environment. Signature matching is accomplished by storing Wi-Fi signatures (signal strengths observed for several detectable access points) for each room and comparing the current signature on the device to stored signatures to find the closest match. In this paper we explain experiments we conducted to explore and optimize Wi-Fi location tracking in one building. While we had hoped for more accurate positioning, we found that only room-level granularity was consistently and reliably achieved. The accuracy of Wi-Fi location tracking is improved as more signature points are stored, but is significantly reduced by the presence of people moving in the area. It also appears that strategically placed access points within a building can contribute to optimum room-level disambiguation of location. Use of a histogram of signal strengths for signatures at a single location may offer a good compromise between a single average and storing a large number of signatures needed for improved accuracy.