Discussions on Data Aggregation in Wireless Sensor Networks

Daisuke Okamura Masaki Bandai Takashi Watanabe Faculty of Informatics, Shizuoka University, Japan Abstract: For energy saving of sensor networks due to power restriction, the data aggregation technique has been proposed. In this paper, we discuss the trade off between communication delay and energy consumption of the technique. For the tandem sensor networks the analytic results as well as simulation results shows that although for the power saving the full aggregation where all data are aggregated is effective, the delay is very large compared with non-aggregation method when a sensing event occurs sparsely in time. Based on the results, to suppress the delay, this paper proposes two partial aggregation methods. One is called RP (random partial aggregation) where some waiting data randomly can be transmitted to the lower node without the sensing data arrival according to the random pushing rate. When the rate equals to zero, the method is equivalent to full data aggregation, while it is non-aggregation when the rate is infinite. The other technique is more sophisticated method called WRP (waterfalls RP). In WRP, each node has its independent random pushing rate. Farther nodes from the sink have larger random pushing rate to suppress the delay. The nodes nearer the sink have less rates and tend to achieve aggregation to suppress the congestion around the sink. The simulation results show the efficiency of WRP.