UC San Diego

This thesis presents the design of a reverse localization system that works to accurately localize the individual antennas of all WiFi access points deployed in a target region. This would enable the localized access points to localize themselves and in turn localize and provide navigation for any new mobile devices entering that network. In order to achieve this, a location-aware mobile robot carrying an antenna array is moved to a number of known locations in the map to collect Channel State Information (CSI) from multiple randomly deployed access points present in the region. The information about the location of the robot and the CSI collected by the robot which by itself acts as a mobile access point is used localize one antenna on the access point. This reverse localization system's tracking algorithm uses phase differences between the localized antenna and the other antennas to identify the distance of separation between the antennas and the orientation of the antenna array present on the access point. Initial results provide centimetre-accuracy for the location of the access point and millimetre-level accuracy for the positions of the other antennas on the AP. This enables us to extend Wi-Fi based localization to various indoor environments.