Energy aware data aggregation in wireless sensor networks

This paper proposes a stream aggregation model that exploits sensors' resources, its implementation and experimental validation targeted to environment monitoring. Instead of sampling and then immediately trasmitting the latest samples towards the sink, a sensor temporarily stores samples in a history located in its RAM. Data is then aggregated and transmitted in any of the following cases: when the history is full, when the size of stored data reaches the optimal packet size or the maximum packet payload, or when a query is received. Since the transmitted packets can reach considerable sizes for a typical WSN, the model also provides an adaptive packet-level FEC technique for maintaining the packet error level below a certain threshold selected by the user. Results show that the proposed technique consistently outperforms the standard sample-and-transmit technique, incrementing the sensors' lifetime of at least 50% while at the same time providing a better performance in terms of network packet loss rate. Source data aggregation can also be combined with other in-network aggregation techniques in order to provide even better results.