Semi-Supervised Top-k Query in Wireless Sensor Networks

摘要

This paper focuses on top-k query in wireless sensor networks and proposes a semi-supervised top-k query approach called CAV. Based on a spanning tree model, CAV adopts histogram technique and an aggregate-verify mechanism to guide query and filter the useless sensing data so as to reduce energy consumption. Besides, this paper further proposes two evaluation schemes of histogram. Compared with existing approaches, CAV is semi-supervised and needn’t set any parameters in advance. Performance analysis and simulation experiment results show the performance of CAV is much superior to that of TAG and it is superior to smooth data distribution than it is to random data distribution.