Research On Decision Level Data Fusion Technology Based On Fuzzy Logic In Wireless Sensor Networks

The wireless sensor networks (WSN) is composed of many low power sensor nodes with wireless transmitter-receiver set; those nodes monitor and collect environmental information, which would be transmitted to base station for processing. The WSN collects large amount of data at a time; the research of how to correctly and efficiently deal with these data is a hot topic of WSN currently. The sensor nodes are usually deployed in harsh environment; the data collected are often inaccurate due to some specific reasons. Therefore, based on such collected data, the users can not correctly judge the current object condition. The decision level data fusion algorithm based on fuzzy logic can solve the problem well.Being set on the background of the National 863 Projects "The Monitoring Technology Analysis of Security Condition for Railway Dangerous Goods Based on Wireless Sensor Networks", this paper summarized the current effective decision level data fusion technology, combined with the development of the security condition monitoring system of transported cotton, then brought up the data fusion algorithm based on fuzzy logic. The algorithm processed and analyzed the collected data to get the accurate condition of the measured object. Main jobs are completed as follows:(1) Analyzed the traditional decision level data fusion algorithms such as adaptive weighted data fusion algorithm and the arithmetic average of the data fusion algorithm; summed up the advantages and disadvantages of both algorithms, as well as the requirements of detection system, therefore further defined the goal of ideal algorithm.(2) Proposed the data fusion algorithm based on the fuzzy logic in two phases. Firstly, used the proximity-based data fusion algorithm to integrate and calibrate similar data, for the purpose of getting more accurate data instead of the poor reliable data with errors. Secondly, used fuzzy reference model to fuse the heterogeneous data from the first phase, got the current condition of the object in supporting the decision-making.(3) The proposed data fusion algorithm simulated in this paper, associated with real measured data, found the conclusion that the proposed method, compared with the traditional fusion algorithm, can more accurately describe the measured object.