Research On Routing Protocol Of WSNs Based On Neural Network

Wireless Sensor Networks(WSNs)face many technical requirements such as real-time routing,node clustering and data integrity.The Artificial Neural Network(ANN)can provide a series of solutions to enhance the dynamic behavior of WSNs to adapt to their surroundings.In this paper,the research focuses on the WSNs routing protocol.Based on the small-scale and large-scale application of WSNs,the paper firstly summarizes the research background and research purposes of the topic,and introduces the WSNs architecture,WSNs and traditional networks and Ad Hoc networks.The principle and algorithm model of ANN are analyzed,and the similarity and functional complementarity between ANN and WSNs are discussed.The performance improvement of ANN on WSNs routing is optimized and experimentally verified.In this paper,the following innovative research findings are completed:(1)A self-organized structure with a simple structure and a small bunched form is proposed.In the ANN-based WSNs routing protocol,in order to keep the plane routing protocol simple,easy to expand,and robust,the network first self-organizes into a smaller bunch in the form of nearest neighbors.Therefore the communication cost is small,and the interference is also small between bunches.(2)With the help of the generalized learning ability of Aptive Linear Neuron(ADALINE),the energy-efficiency,network robustness,traffic reduction,and data are learned through data relationship learning by means of mutual cooperation and supervision of neighboring "node pairs".In addition,the energy of the tightly distributed nodes is preferentially consumed,and the equilibrium distribution of the surviving nodes is better maintained in the case of a large number of nodes dying,and verified by experiments.(3)Based on the large-scale application of WSNs,a routing model based on the ability of convolutional neural network(CNN)analysis to improve the performance of WSNs is proposed.The model forms a corresponding relationship between CNN nodes and WSNs nodes,and implements CNN network model in parallel within WSNs.It increases network robustness by using the peer relationship between CNN and WSNs distributed storage.Using WSNs' distributed processing capability to implement local decisions with CNN.It has the effect of reducing network traffic and increasing the practicability and feasibility of large-scale WSNs.Finally,the flow and energy consumption are analyzed in the network through simulation experiments.The simulation results show that the network traffic is significantly reduced both at the cluster head and the base station level,and the energy consumption is significantly reduced after the model is compressed.Figure[27] table[2] reference[73]...