Abstract:

The modelbased data collection technology which can inhibit unnecessary data transportation and save energy effectively has been widely applied in sensor networks. We improve the traditional modelbased data collection technology and put forth an approximate data collection algorithm based on the Kalman filter,called ADCA, which can collect data effectively within a given range of error.In ADCA,sensor nodes are organized as clusters.Cluster header and cluster members build the Kalman filter respectively and save their mirror models.Cluster header can produce approximate data for cluster members,so some user queries can be answered by cluster header.Experiments show that ADCA has good performance.

Key words: prediction model;sensor network;approximate algorithm;data collection