Dynamic data aggregation and transport in wireless sensor networks

Abstract

In wireless sensor networks, in-network aggregation is the process of compressing locally the data gathered by the sensor nodes, so that only the compressed data travel across several hops to their destination. We address the problem of aggregating data generated by sporadic events in random locations of the monitored area. The sensor nodes keep their transceivers off most of the time in order to preserve their batteries, and these sleep periods dominate the time to react to the events. We propose a distributed protocol that, after each event, constructs a routing tree to regulate the aggregation process. It is cross-layer because, in order to accelerate the tree construction process, the routing decision considers the sleep periods of the nodes. If the nodes sleep for long periods, our protocol divides the tree construction time by the number of hops when compared to centralized protocols. For a fixed maximum tolerable delay, this allows us to extend the sleep periods and thus to save energy. Our simulations reveal that this comes at the price of an aggregation tree with degraded performance, but we retain an advantage over trees not customized to each event. Our protocol requires global time synchronization and periodic link state monitoring, especially as the network size increases.