Data pre-forwarding for opportunistic data collection in wireless sensor networks

In many potential wireless sensor network applications, the cost of the base station infrastructure can be prohibitive. Instead, we consider the use of mobile devices carried by people in their daily life to collect sensor reports opportunistically. Considering that human mobility demonstrates strong spatial locality and sensor nodes need to be deeply duty-cycled for longevity, data pre-forwarding (DPF), in which sensor reports are forwarded to sensor nodes visited by people more frequently even though they are not currently being visited, should be a promising scheme to improve the performance of opportunistic data collection. In this paper, a distributed DPF mechanism is proposed to exploit the spatial locality of human mobility in the context of opportunistic data collection. The communication protocol is first carefully designed so that sensor nodes could rendezvous and communicate with their neighbors and mobile nodes energy efficiently. A simple heuristic algorithm is then designed so that sensor nodes could decide the number of sensor reports exchanged with their direct neighbors for improving network throughput under the energy constraint of sensor nodes. The distributed DPF mechanism has been implemented in Contiki-OS and evaluated with Cooja. Evaluation results indicate that this proposal significantly outperforms the default approach (with no data pre-forwarding).