The Research Of Data Fusion Algorithm In Sensor Network

Sensor networks have been widely used in military, industrial, agricultural and commercial areas. In order to ensure coverage of sensor networks, sensor networks will be arranged in a large number of redundant nodes. Data fusion can effectively reduce the amount of data transceiver, reduce node energy consumption and prolong the network life. So data fusion is a very important role in sensor networks. The algorithm is the main content of the data fusion.This article attempts to send the sensor data types at different times as a rough set of sample space. Use typical sensor data as rough sets condition attribute. Use a sensor on behalf of a group sensors in some time as a rough set of decision attribute. Cluster them in kohonen neural network. The number of cluster centers is the distribution of the number of fusion. Eventually get a rough set-kohonen algorithm.Base on the above algorithm, this article try to add topological structure to the granular space. Make rough set theory transited to quotient space theory. Then get a new algorithm: quotient space-kohonen algorithm. Add a data fusion layer to the real network to support the topology. Made this method successfully applied to the sensor network.Then describe the calculation of the computing with words algorithm. Computing with words can support fuzzy granular which is different form the above two algorithms. So we don't need to add another fuzzy neural network to data fuzzy with the computing with words algorithm. This article gives an example to demonstrate the operation process of computing with words. Use the initial sensor measurement data as an initial state. Final state as a terminate state, And form a transition matrix. Finally completed use the computing with words algorithm in sensor network.Experiments show that the quotient space-kohonen algorithm has faster speed than rough set because it adds structure. However it also has less number of cluster distributions than rough set-kohonen algorithm. The computing with words fusion can quickly get some of the sensor cluster distribution by control the number of the result.