Fine-grained in-door localisation with wireless sensor networks

Many WSN algorithms and applications are based on knowledge regarding the position of nodes inside the network area. However, the solution of using GPS based modules in order to perform localization in WSNs is a rather expensive solution and in the case of indoor applications, such as smart buildings, is also not applicable. Therefore, several techniques have been studied in order to perform relative localization in WSNs; that is, to compute the position of a node inside the network area relatively to the position of other nodes. Many such techniques are based on indicators like the Radio Signal Strength Indicator (RSSI) and the Link Quality Indicator (LQI). These techniques are based on the assumption that there is strong correlation between the Euclidian distance of the communicating motes and these indicators. Therefore, high values of RSSI and LQI should indicate physical proximity of two communicating nodes. However, these indicators do not depend solely on distance. Physical obstacles, ambient electromagnetic noise and interferences from other wireless transmissions also affect the quality of wireless communication in a stochastic way. In this paper we propose, implement, experimentally fine tune and evaluate a localization algorithm that exploits the stochastic nature of interferences during wireless communications in order to perform localization in WSNs. Our algorithm is particularly designed for in-door localisation of moving people in smart buildings. The localisation achieved is fine-grained, i.e. the position of the target mote is successfully computed with approximately one meter accuracy. This fine-grained localisation can be used by smart Building Management Systems in many applications such as room adaptation to presence. In our scenario, our proposed algorithm is used by a smart room in order to localise the position of people inside the room and adapt room illumination accordingly.