A model-based WiFi localization method

Due to the proliferation of WiFi access points, indoor localization methods based on WiFi signal strengths are becoming more and more attractive because they don't require additional infrastructural costs beyond the existing WiFi infrastructure. Many research projects were proposed based on this approach, but most of them only focused on the processing of the signal strength data to obtain the user locations. Very little attention was paid to the mobility patterns of the users. In this paper, we propose a method based on (i) knowing the floor model, (ii) continual tracking of user locations and (iii) back-tracking from the current location to previous locations to resolve localization ambiguities. We implemented the system in a life environment and performed experiments to measure the localization accuracies. We found that our method identified all test paths accurately with the exception of a challenging case where two locations were connected together with a thin wall. We discuss ways to handle this situation.