Efficient data aggregation and transport in wireless sensor networks

We consider the problem of reporting events using wireless sensor networks. To reduce the data volume generated by each event, the correlated data from the (neighboring) nodes that detect the event must be brought together to be processed and compressed before relaying the result across several hops to the data sink. This process is supported by an aggregation tree, which specifies the flow of information towards the sink. For efficiency, aggregation trees should compress the data close to their sources. We propose two solutions to the event-triggered reporting problem. Firstly, we propose the first protocol to use a staggered schedule in the construction of the aggregation tree. Due to the use of such a schedule, our protocol divides the tree-construction time by roughly the number of hops in the network, and this advantage comes only at the expense of a small degradation of the quality of the obtained aggregation tree. Secondly, we consider a multi-hop cluster-based topology with fixed aggregation points. This topology is appropriate for large networks with unreliable radio links. We approximate the optimal cluster size distribution and evaluate the improvement over a uniform cluster size distribution. Copyright © 2009 John Wiley & Sons, Ltd.