Study Of Data Fusion Algorithm Based On Temporal-spatial Correlation In Wireless Sensor Networks

Wireless Sensor Network(WSN)as a new mode of data acquisition and processing,has been widely used in military aviation,environmental monitoring,health care and other fields.However,WSN nodes are distributed densely and sample frequently.The temporal-spatial correlation between sensed data caused redundant data.Huge amounts of redundant data transmission bring huge pressure to the restricted energy,storage capacity and network bandwidth of WSN.Data fusion can efficiently reduce the redundant transmission data,improve the efficiency and accuracy of collection data.Therefore,the study on WSN data aggregation has a very important academic significance and engineering value.In this paper,the main work is divided into the following three points:Firstly,Respectively from two aspects:with the data correlation and with routing protocol,the paper reviews the research of data fusion at home and abroad.Based on data correlation,data fusion mechanism against can be divided into temporal correlation data fusion algorithm,spatial data fusion algorithm and spatial-temporal correlation data fusion algorithm.Based on routing protocol,data fusion mechanism can be divided into query routing data fusion protocols and hierarchical network data fusion routing protocol.Secondly,in the process of solving the signal WSN node data fusion,specific to the phenomenon that sensor data has long-term trend steadily and instantaneous fluctuation strongly,the paper proposes a temporal correlation data acquisition algorithm based on double model driven.First of all,through the analysis of the time series,a model of PLR is set up,to describe the long-term variation trend.Secondly,through the analysis of residual error series between the predicted value by trend model and the observed value,a model of AR is set up,to describe the Instantaneous fluctuation of object.Finally,the two model work concurrently,detecting the validity of the model in real time and updating model parameters dynamically.Simulation and experimental results show that,for different data rate,the algorithm can effectively reduce the amount of data transmission,fitting precision is ensured synchronously.Thirdly,based on the double model above,the paper proposes a spatial correlation secondly clustering algorithm.Through clustering twice,this algorithm divides WSN into three layers of structure: the first clustering divided all of WSN nodes into multiple clusters,this step executives by the classical clustering algorithms,mainly based on geographic location and residual energy.The second clustering further divides the cluster into multiple sub clusters,primarily based on trend model similarity and adjusting model similarity between different nodes.Finally,each cluster consists of representative nodes,the redundancy nodes and abnormal nodes.The representative nodes represent the sub-cluster and abnormal nodes represent themselves to monitoring and reporting information.Redundancy nodes sleep.Simulation and experimental results show that the algorithm reduces amount of data transmission,improves the accuracy of data collection and prolongs the life of network.