Robust and Dynamic Data Aggregation in Wireless Sensor Networks: a Cross-layer Approach

In-network data aggregation is an effective technique to reduce communication cost in wireless sensor networks. Recent works have focused on two issues individually: dynamic aggregation to handle irregular traffic of events and robust aggregation to tolerate packet losses. However, how to achieve both the objectives simultaneously is still not touched. In this paper, we propose a cross-layer approach to robust and dynamic data aggregation by making use of direct support from MAC layer. A new MAC protocol, DA-MAC is delicately designed to serve such purpose. Based on the channel contention information obtained from DA-MAC, a node can dynamically determine where and when to do aggregation. To cope with packet losses, a virtual overlay, Rings is adopted to forward one packet to multiple nodes. We have implemented our design in TinyOS based sensor networks. Performance evaluations though simulations and experiments show that, compared with existing algorithms, our proposed solution is more efficient in terms of both time and energy cost.