Sensing in Rich Bluetooth Environments

The abundance of short-range radios, such as Bluetooth, in the current office and urban environments opens up new possibilities for data modeling and analysis. Although individual devices are constantly moving and active only sporadically, various environments have specific statistical characteristics. Mobile handsets are often personal in nature, so the statistics reflect also behavioral patterns of their users. In this paper we first demonstrate how prototypical patterns of behavior may be found in data produced by Bluetooth scanning. It appears that cyclical nature of the patterns reflect well various daily routines of test persons. Our second model demonstrates that rich Bluetooth environments are often stable enough so that they can be used for locationing without any base stations.