A cross pruning framework for Top-k data collection in wireless sensor networks

Energy conservation is a key issue for algorithm designs in wireless sensor networks. In this paper, we explore in-network aggregation techniques for answering top-k queries in wireless sensor networks. A top-κ query retrieves the κ data objects with the highest scores evaluated by a scoring function on interested features of sensor readings. Our study shows that existing techniques for processing top-κ query, e.g., Tiny AGgregation Service (TAG), are not energy efficient due to deficiencies in their routing structures and data aggregation mechanisms. To address these deficiencies, we propose to develop a new cross pruning (XP) aggregation framework for top-κ data collection in wireless sensor networks. The XP framework incorporates several novel ideas to facilitate efficient in-network aggregation and filtering, including 1) building a cluster-tree routing structure to aggregate more objects locally; 2) adopting a broadcastthen-filter approach for efficiently suppressing redundant data transmissions; and 3) providing a cross pruning technique to enhance in-network filtering effectiveness. An extensive set of experiments based on simulation has been conducted to evaluate the performance of TAG and the proposed XP framework. The experimental results validate our proposals and show that XP significantly outperforms TAG in energy cost.