Research On Data Aggregation Of Wireless Sensor Networks Based On The Neural Networks

Wireless Sensor Networks (WSN) is a new intelligent information network system for real-time monitoring, which integrates the sensor technology, network technology, embedded control technology, distributed computing and storage technology and wireless communication and so on. It becomes one of the focus fields of the current computer network research. The sensor node is distributed intensively in WSN, the adjacent nodes collect high similar data, the data of network transmission has some redundancy, the communication energy consumed too much.Related studies have shown that the most of the energy consumption is consumed in the data transmission, the data fusion technology can effectively reduce the redundant information of network, to improve the efficiency of data transmission and save the energy.At present, many scholars research data fusion in the network layer and application layer of WSN. A new data fusion algorithm of WSN based on the neural networks (Back-Propagation Data Fusion Algorithm, for short as BPDFA) is proposed, which combines BP neural network and the hierarchical routing protocol of WSN.Firstly, the hierarchical routing protocol LEACH of WSN is analyzed, which includes network model and working procedures. In the view of total energy consumption, the calculating method of the optimal cluster head number is proposed, to analyze the advantages and disadvantages of LEACH.Secondly, an improving routing protocol LEACH-E is proposed, which solves cluster head uneven distribution, cluster head election did not consider the node residual energy, LEACH-E considers the number of adjacent nodes in the clustering and the residual energy of the node, to balance whole network load.Thirdly, BPDFA algorithm model integrates the BP neural network algorithm into the improved routing protocol LEACH-E. The algorithm model puts the hierarchical structure of BP neural network into clustering routing protocol, which designs a three-layer BP neural network model in each cluster structure. The mass raw data collected are processed cross the model, and then the represented features processed data will be transmitted to the sink node. So the transmission amount of data information is reduced, to save the energy and prolong the network life time.Finally, the algorithm is simulated using NS-2.The simulation results show that compared with LEACH, BPDFA algorithm has a better performance in balancing node energy and prolonging the lifetime of the network.