Dynamic Wi-Fi fingerprinting indoor positioning system

A 2014 IPIN conference paper proposes indoor positioning from dynamic Wi-Fi fingerprints. It can locate people where no wireless network is deployed or where the network has been badly damaged.

Omar Costilla-Reyes and Kamesh Namuduri presented the paper at the Fifth International Conference on Indoor Positioning and Indoor Navigation (IPIN 2014), held in Busan, South Korea, in October 2014. The paper is drawn from Dr. Costilla Reyes’s master’s thesis in electrical engineering at the University of North Texas, which addressed dynamic indoor positioning with wireless sensor networks and machine learning.

Satellite navigation is often unavailable or unreliable inside buildings, so indoor positioning systems estimate location from other signals. Wi-Fi fingerprinting does this by comparing the signal strengths a device observes from nearby access points against a reference map of measurements taken at known positions. The paper proposes improving such systems with dynamic access points and fingerprints (DAFs), and further proposes a positioning system that relies solely on DAFs, which uses Wi-Fi signals from devices carried by people in the building. The walking pattern of indoor users is classified as dynamic or static for positioning purposes, and the authors show that a conventional system built on static fingerprints improves when dynamic fingerprints and access points are considered. Accuracy is evaluated with several positioning algorithms and an access-point selection strategy.

Because the approach does not depend on fixed infrastructure, it can locate people where no wireless local area network has been deployed or where the network has been badly damaged, for example after a natural disaster, which makes it relevant to search-and-rescue operations and to extending the coverage of an existing indoor positioning system. The paper complements the indoor SLAM project Dr. Costilla Reyes carried out at UNT in the same period, which approached indoor localization with a Kinect depth camera rather than wireless signals. SLAM, simultaneous localization and mapping, builds a map of a space while tracking where the sensor is within it.