Research Of The Data Fusion Technology In Wireless Sensor Networks

The energy of Wireless Sensor Network nodes is limited and cannot beresupplied once the WSN is deployed. The data collected by nearby nodes contains alot of redundancy information. Hence, forwarding these redundant data to the sinknode will lead to unnecessary energy consumption. Data fusion technology caneffectively reduce redundant data forwarding, achieving the purpose of extendingthe life cycle of networks. Combining data fusion technology and routing protocolsin the network layer of WSN can dramatically reduce redundant data beingforwarded. And meanwhile, it achieves a higher degree of integration.This thesis focuses on data fusion in network layer. Based on the analysis andstudy of LEACH routing protocols as well as classic data fusion algorithm, thisthesis introduces an improved hierarchical fusion algorithm. Our research is basedon a hierarchical routing LEACH protocol, since it has no specific integration policyissues. Following the approach in the LEACH protocol hierarchy, the algorithm isdivided into three levels of integration policy. Because the data collected by normalsensor nodes is inaccurate, this thesis uses Kalman filtering to treat raw data toimprove data accuracy, and then uses simple threshold estimates mechanism toinhibit the forwarding of redundant data. At the cluster node, since data collectedform different sensor nodes may contain many errors, this thesis will use theGeluobusi guidelines to check data consistency, removing abnormal data in order toimprove the accuracy of the integration. And then, we use consistency of fusionalgorithm for homogeneous data fusion to remove redundant data and improve theaccuracy of fusion. At the sink node, the user needs to use different types of data tomake a final decision,and this thesis makes decision fusion by using fuzzycomprehensive. Finally, MATLAB is used to simulate and verify the validity of thisalgorithm.The proposed algorithm can reduce the redundant data forwarding, avoidabnormal data to poison results, to reduce the node energy consumption and improvethe accuracy of data collection.