Exploring data forwarding with Bluetooth for participatory crowd monitoring

Participatory crowd monitoring estimates the size and dynamics of a crowd based on position data shared by people in the crowd. At large-scale events, these small but frequent uploads compete for bandwidth with the crowd's social networking activities causing transmission errors and long delays. Forwarding collected data within a peer-to-peer network allows for bundling that data and assigning the upload task to a few selected peers. On modern phones, however, peer discovery and networking is challenging due to tight restrictions of mobile operating systems. This paper shows how the announcement of internal state transitions via Bluetooth enables peers to align their operations and to accommodate for issues such as bandwidth restrictions, inexact timing, limited background activity, and idle times. Experiments on Android phones show that already for a small set of phones the proposed protocol works reliably and significantly reduces the use of the networking infrastructure. With the data delay and energy consumption in acceptable bounds, the protocol presents a viable solution for when network overload puts crowd monitoring data at risk of not being delivered in a timely manner.