Automatic Localization of Passive Infra-Red Binary Sensors in Home: from Dense to Scattered Network

Abstract : Location of residents in a household is one of the critical information to provide context-aware services. Passive Infra-Red (PIR) binary motion sensors have become the de facto standard technology used in the home by tracking systems due to their low energy consumption and their wide range of coverage. However, installing and managing this network of PIR sensors is difficult for typical residents, such as older adults, with low technical skill. To enable easy deployment of such a system by anybody, we present an extension of a method to automatically identify the location of multiple PIR sensors in a house from the observed motion detection event sequences. Thanks to a floor plan given as prior knowledge, the method estimates the distance between pairs of sensors and identifies particular patterns in the observations to predict the rooms where those sensors are most likely located. The method, which was designed to deal with dense sensors network is adapted to the case of scattered sensors which correspond to most traditional houses. Experimental results on a realistic home show that our method can estimate the location of sensors placed close to the anchor locations with only a few confusions. The experiments also revealed challenges to be addressed to make this method scale to various house configurations.
Document type :
Conference papers
Complete list of metadatas

Cited literature [14 references]  Display  Hide  Download
Contributor : François Portet <>
Submitted on : Wednesday, June 26, 2019 - 6:35:23 AM
Last modification on : Tuesday, August 6, 2019 - 7:46:49 AM


Files produced by the author(s)





Nathavuth Kitbutrawat, François Portet, Hirozumi Yamaguchi, Teruo Higashino. Automatic Localization of Passive Infra-Red Binary Sensors in Home: from Dense to Scattered Network. 2019 IEEE Conf on Pervasive Intelligence and Computing, Aug 2019, Fukuoka, Japan. pp.848--853, ⟨10.1109/DASC/PiCom/CBDCom/CyberSciTech.2019.00154⟩. ⟨hal-02165530⟩



Record views


Files downloads