Location: Eli Lilly Room (6106) - Sennot Square

Abstract:

Wireless Sensor Networks have dismal packet delivery, and it becomes worse with an increase in the number of nodes and data rate in the network. This is a pity because numerous applications where a wired network cannot be used (e.g. battlefield surveillance) or applications that still require a large wired network infrastructure (e.g. structural health monitoring) would otherwise benefit from wireless sensor networks, in terms of cost, ease of installation, maintenance, and large scale operation.

Following locally available knowledge at a wireless sensor node to determine what data delivery rate to use for transmission can dramatically overcome this obstacle. We propose such a mechanism in this paper. We show that using this mechanism, the packet delivery success would increase manifold. It also enables us to tune the network to achieve a variety of desired performance targets.