Mobile Wireless Middleware, Operating Systems, and Applications. Third International Conference, Mobilware 2010, Chicago, IL, USA, June 30 - July 2, 2010. Revised Selected Papers

Research Article

Location Cognition for Wireless Systems: Classification with Confidence

Download47 downloads
  • @INPROCEEDINGS{10.1007/978-3-642-17758-3_11,
        author={Stefan Aust and Tetsuya Ito and Peter Davis},
        title={Location Cognition for Wireless Systems: Classification with Confidence},
        proceedings={Mobile Wireless Middleware, Operating Systems, and Applications. Third International Conference, Mobilware 2010, Chicago, IL, USA, June 30 - July 2, 2010. Revised Selected Papers},
        proceedings_a={MOBILWARE},
        year={2012},
        month={10},
        keywords={monitoring statistics fingerprinting classification location cognition confidence cognitive radio systems},
        doi={10.1007/978-3-642-17758-3_11}
    }
    
  • Stefan Aust
    Tetsuya Ito
    Peter Davis
    Year: 2012
    Location Cognition for Wireless Systems: Classification with Confidence
    MOBILWARE
    Springer
    DOI: 10.1007/978-3-642-17758-3_11
Stefan Aust1,*, Tetsuya Ito1,*, Peter Davis2,*
  • 1: NEC Communication Systems, Ltd.
  • 2: Telecognix Corporation
*Contact email: aust.st@ncos.nec.co.jp, ito.tts@ncos.nec.co.jp, davis@telecognix.com

Abstract

Location cognition is a challenging task in cognitive wireless systems when there is no explicit location information system available, such as Global Positioning System (GPS) or dense wireless beacons. This paper decribes a simple-but-effective method of real-time location cognition which can be used by wireless devices in WLAN systems without depending on any location service infrastructure. The method is based on monitoring, learning and recognizing the statistics of received data traffic, with an awareness of the confidence in the recognition result. It uses the property that traffic statistics such as average and variance of throughput are correlated with the location of the transmission. Locations are recognized by comparing monitored statistics with a set of reference distributions and identifying the best match. A measure of the confidence in the location classification result is obtained by comparing matches with multiple candidate locations. It is demonstrated that the method can be implemented as middleware for use with WLAN devices and used to recognize multiple locations, indoor and outdoor. It is also demonstrated that the method can be used to detect the distance between a sender and receiver.