Research On A Support Function Improvement And Its Application For WSN Data Fusion

Wireless sensor networks(WSN) nodes are susceptible to environmental factors, so the data collected by them may be abnormal. Currently, the main way of data fusion techniques performing data fusion is that the selected data were directly sent to the fusion node by source node. However, in this way, the abnormal data would seriously affect the final fusion result, moreover, it would produce more traffic which is not conducive to network energy saving.This paper compared and analyzed some kind of major methods about wireless sensor networks data fusion technology, chose the data fusion algorithm based on support function to study, and proposed an improved support function to identify abnormal data which improved the accuracy of data fusion. And then combined the fusion algorithm based on improved support function with an initial fusion algorithm to put forward a new twice fusion module. The module can reduce energy consumption of WSN when guaranteeing the accuracy of the integration. The main research work included in the papers are as follows:① For the problem that the abnormal data will affect the fusion results, this paper proposed an improved support function for data fusion. Combined the concept of self-support and gray proximity theory to improve the Exponential Decay support function. Used the improved support function to compute the fusion results. The algorithm considers the link between the data more fully, and can discern the abnormal data effectively. When performing data fusion, adding to the abnormal data smaller weights to reduce its impact on the fusion results and improve the accuracy of the fusion. The simulation results of Matlab2013 show that: compared with the weighted data fusion algorithm and the algorithm based on exponential decay support function, the algorithm supported by this paper can improve the accuracy. When adding some abnormal data into the experiment data, this paper’s algorithm can discern the abnormal data effectively and reduce their impact on the fusion results.② For the problem that the nodes directly sent the data to the fusion node would produce more traffic and consume more energy, WSN nodes fuse the data before sending them to fusion node can reduce the amount of data and the energy consumption of WSN. Combine this idea with the fusion algorithm based on improved support function to propose a twice fusion module. This module takes the accuracy of data fusion and WSN energy efficiency into account. The simulation results of Matlab2013 show that the twice fusion module in this paper can reduce energy consumption of WSN when guaranteeing the accuracy of the integration.