Automatic Site Survey for RSS-Based Wi-Fi Positioning Using Unlabeled Data

Preparing a database of Wi-Fi access points (APs) is essential for deploying Wi-Fi-based positioning solutions. This study proposes an automated method to discover APs installed in an indoor site by exploiting unlabeled data provided by mobile users. Furthermore, the positions of these APs are estimated by comparing the shapes of two trajectories, which are obtained using the provided Wi-Fi and sensor data. The effectiveness of the proposed method is verified with excessive experimental campaigns, including an extreme case scenario, where the positions of 60 APs are obtained from only three APs with known positions.