Robust Aggregation Algorithm in Sensor Networks

Abstract

Saving energy to prolong network life is a big challenge for WSNs (wireless sensor networks) research. In-Network query can reduce the number or size of packets through processing data in intermediate nodes so as to consume energy effectively. Present aggregation algorithms suppose all the sample data are correct. The existing outlier detection algorithms regard detection rate as the primary object and do not consider energy consumption and query characteristic. So the simple combination of the two aspects can not bring good performance. By analyzing the influence of faulty and outlier readings to aggregation results, this paper puts forward a robust aggregation algorithm RAA (robust aggregation algorithm). RAA improves traditional aggregation query using reading vector to judge whether a faulty or outlier has happened. RAA deletes faulty readings, aggregates normal readings and reports outliers. Thus, customers can know the networks condition clearly. Finally, this paper compares RAA and TAGVoting which uses tiny aggregation algorithm to complete aggregation and the Voting algorithm to realize outlier detection at the same time. Experimental results show that RAA outperforms TAGVoting in terms of both energy consumption and detection rate.