Outdoor Location Estimation in Changeable Environments

One approach to location estimation constructs a radio map of received signal strength (RSS) measurements at different known locations. However, location-based systems that depend on RSS alone are susceptible to inaccuracies caused by several factors, such as changes in humidity, temperature, the number of users and the physical environment. In this paper, we present a novel algorithm in which one radio map is generated as a reference map and adjusted for different run-time environmental conditions. A small sample of new RSS samples for the new environment is collected, with locations, and used to build a model to calibrate new measurements to the reference radio map. The calibration is not uniform and depends on the observed RSS. The effectiveness of the proposed method is demonstrated using real GSM data sets collected from a three-day music festival in London Victoria Park. Results are presented with and without applying the correction. A state-of-the-art cluster-based deterministic location estimation algorithm is used throughout.