Research On Application Of Compressed Sensing In Data Fusion Of Wireless Sensor Networks

Compared with the general wireless networks,wireless sensor networks(WSNs)is a distributed network with a large number of nodes and energy constraints.The data fusion algorithm,which can effectively remove invalid or redundant information,collected and transmitted by WSNs,is one of the main research points of wireless sensor network.Compressed sensing theory is a new data processing method,which has been used in more and more fields in recent years.Based on Compressed sensing theory and combined with the network structure of wireless sensor network,the sparse projection mechanism and distributed compressed sensing algorithm are improved to increase the data fusion efficiency of wireless sensor networks.Specific research contents are as follows:In this thesis,a dynamic data fusion tree algorithm based on compressed sensing theory is proposed to solve the problem that the existing sparse projection algorithm uses the static shortest path algorithm to calculate the path that the ordinary node transmits data to the sink node,which leads to the high efficiency of data fusion and the excessive overhead.Firstly,this thesis generates the sparse matrix based on the sparse projection mechanism and selects the nodes that need to be involved in the projection according to the sparse matrix.Secondly,when the projection is switched,the data fusion tree is dynamically generated according to the minimum cost principle according to the departure and join of the projection node,which improves the efficiency of data fusion and avoids the repeated calculation of the path for the node.Simulation results show that the dynamic data fusion tree algorithm reduces the cost of data transmitting of wireless sensor network under the premise of ensuring the precision of signal reconstruction.Based on the distributed source coding and the first joint sparse model,a distributed compressed sensing algorithm based on edge information is proposed.This thesis improves the data fusion algorithm of wireless sensor networks based on data correlation is only used to design the spatial or temporal correlation of data.Using the temporal and spatial correlation of the data collected by the nodes,the data fusion is carried out at the cluster head.The fusion method is that the cluster head can use the data of one node as the side information which carry all the information of the signal,and the data of the other nodes only need to carry the information of the independent sparse part,after receiving the data transmitted by the member node.Simulation results show that the distributed compressed sensing algorithm based on edge information can more effectively fuse the signal conforming to the first joint sparse model,and reduce the data transmission and transmission energy of the network.