Energy-efficient and Balanced Top-k Query Techniques in Sensor Networks

Energy efficiency and balance of sensor nodes in processing top-k queries can prolong the lifetime of wireless sensor networks. In this paper, an Energy-efficient and Balanced query Sampling Top-k algorithm named EBSTopk(,) is proposed, which is based on the sampling techniques and the spatial correlations among sensor nodes. First, the sensor network is partitioned into several regions. Next, the linear regression prediction model and Gaussian prediction model are constructed based on the spatial correlations of pairwise sensor nodes. Then, the criteria of high spatial correlation is established due to the given relative error bound and the confidence level 1-. Finally, according to the predicting models and criteria above, two energy balanced algorithms named EBSTopk(,)-LR and EBSTopk(,)-MG are proposed, which are based on iterative random sampling technique. Experimental results show that, the proposed EBSTopk(,) algorithms not only reduce the global energy consumption in wireless sensor networks, but also achieve balanced energy consumption among all sensor nodes after continuous processing top-k queries.