Lessons Learned from Bluetooth/Wifi Scanning Deployment in University Campus

This paper presents the detailed design and implementation of the joint Bluetooth/Wifi scanning framework called UIM 1 , which collects both location information and ad hoc contact of the human movement at the University of Illinois campus using Google Android phones. In particular, we present the architecture of UIM and how its sub components interact to obtain the performance reliability as well as conserve phone battery for the prolonged experiment period. With the movement trace collected by UIM, we first present the findings about number of scanned devices, types of collected devices, and instant cluster size distribution. Then, we study the two graphs formed by the ad hoc trace including connectivity graph and contact graph. We find that the former exhibits a small-world network in structure while the node degree distribution of the latter exhibits an ExponentialZipf distribution. Finally, we present a novel and efficient algorithm called UIM Clustering to cluster collected wifi access points into clusters and use these clusters to represent locations. Our analysis shows that the distribution of number of locations visited by experiment participants can be fitted by an exponential function.