A Research On Sampling Mechanism Based On Spatial-Temporal Correlation In Wireless Sensor Networks

Reducing the energy consumption is one of the key points of the research and design of wireless sensor networks. Since the sensor nodes have been densely deployed in the monitoring field, the sensing data collected by the nodes presented highly spatial-temporal correlation and redundancy. Based on the spatial-temporal correlation of the sensing data, this paper proceeded the research on sampling mechanism. Our purpose is to depress the transmission of redundant data, save the energy of wireless sensor networks. The research comes with the three following parts:Based on the spatial-temporal correlation, we discussed the current situation of sampling mechanism in wireless sensor networks at home and abroad, then classified the available mechanism: the mechanism based on centralized method and the mechanism based on vertical data comparison method.Wireless sensor networks would produce a lot of abnormity data as a result of environment noise, loss of energy and hardware failure. The data with errors will have a disgusting influence on accuracy of the detection results. To solve the problems mentioned above, a mechanism for abnormity event detection based on the spatial-temporal correlation is proposed. By using the correlation of sensing data, the mechanism can eliminate the abnormity data caused by the failure of nodes. Group of nodes can be divided into several teams working in turn to save energy besides performance of detection. During the process of abnormity event judgment, the mechanism shorten the time it took by raising the sampling rate to insure the timeliness.For the applications which the sensing data presented certain periodicity, we propose a sampling mechanism of sensors network using the autonomous partition of the interval based on linear regression model. In our proposed scheme, the interval can be partitioned independently based on the fluctuation of data flow, and then assigns appropriate sampling rate for each interval. Simulation shows that our scheme can further utilize the spatial-temporal correlation of sensing data. Our scheme can also make the utilization of energy reasonable and capture fluctuations of data accurately.