A RF map-based localization algorithm for indoor environments

The paper presents a localization algorithm for Wireless Sensor Network (WSN) units which estimates the spatial position of a probe unit by exploiting received signal strength information. The algorithm is particularly suitable for indoor applications where the presence of furniture, objects, walls and the induced reflection effects create a strongly non linear attenuation field map which makes ineffective the use of traditional 1/r(k) theoretical signal attenuation models. The field map is learned from samples acquired during a preliminary environment exploration campaign. The developed algorithm performances stay at the top of its class; we experimentally verified that it is able to locate nodes with an accuracy of 1.2m and a confidence of 50% in MICA2 WSN.