A Survey Of Data Fusion In Wireless Sensor Network

The wireless sensor network is composed by many sensor nodes which have the low power wireless transmitter-receiver set, they collect environment information from the different environmental monitoring and transmit it to the distant base station to carry on processing, and sensor network research starts to US 1990s. But from before always puzzles the people to the present is the sensor nodes' energy problem. Because the sensor nodes generally all are the battery power supply, and the deployment is very broad, then replacing the battery energy to lengthen the network life cycle is complex, therefore how to save the energy is the hot spot which at present studies. In order to avoid wasting the correspondence band width and the energy as well as reducing the efficiency of the collection of information, the data fusion technology played the very major role.From the end of 1990s, the information fusion in domestic has developed the general character essential technology which has been paid attention in every way, many scholars devote to the research of the maneuvering target track, distributional examination fusion, distributional information fusion, target identification and decision information fusion and multi- sensors synthesis track and localization. The data fusion technology may realize in sensor network protocol stack many levels, also may realize in the MAC protocol, also may realize in the route protocol or in the application level protocol. Thus according to the comparison, the present paper center aims at the data fusion for the application level data.Induct the fusion method in the single sensor system, according to the limited resource; eliminating the uncertainty in the survey may obtain the more accurate reliable measurement result, in order to enhance the data processing the precision. Regarding the system is the linear model and the system and the sensor noise is the Gaussian distribution white noise, and the object to deal is the redundant sensor information in the dynamic environment. According to the application layer data, the article has selected batch estimate fusion algorithm based on the Kalman filter, this algorithm mainly carries on the sensor information which are the same kind for batch estimate, through a series of weighting formula to make the calculated result to approach the true value.Main jobs completed are as follows:(1) Compared with other traditional data fusion methods, the batch estimate arithmetic is proposed based on Kalman Filter.(2) Compared four data fusion arithmetic with reference as well as emulated a large number of data from accuracy, complexity and energy dissipation and obtain advantages and disadvantages of each one.(3) Applied the batch estimate arithmetic to granary monitor device and provide relevant hardware configuration.Experiment result indicates: the accuracy of batch estimate arithmetic based on Kalman Filter merged under valid node is much higher than other arithmetic.