Abstract:

We show that it is possible to extract geometric descriptions of the spaces observed by sensor networks, even if the network consists of sensors that are of very limited ability:  such as motion detectors. By using statistical techniques and relying only on the unconstrained patterns generated by the occupants of the building we show how to recover information about sensor geometry. This is important to the ubiquitous computing community since ubiquitous sensors and the context that the provide will only become a reality if the sensors are cheap, low-power, and sel-configuring.