Research On Improving WSNs Clustering Routing Protocol Based On Clustering And Neural Network

Wireless Sensor Networks(Wireless Sensor Networks,WSNs),as an important "nervous tip" of the Internet of Things(IoT),have now become a representative of a new generation of efficient information acquisition and processing technology.It uses sensors that are randomly deployed inside the network.Nodes perceive the surrounding environment to complete the monitoring tasks in the application.As WSNs show their talents in different fields,defects such as weak node processing capabilities and low energy efficiency have gradually become prominent.As an efficient routing protocol,the clustering routing protocol has the characteristics of low energy consumption,low latency,and high expansion,and it performs well in balancing the energy distribution of WSNs nodes.To solve the problem of the uneven load caused by the random selection of cluster heads in traditional cluster routing,this thesis improves the process of clustering and cluster head selection based on the research of the LEACH clustering routing algorithm.Taking into account the influence of node distribution under different scene scales,this thesis proposes two different improvement schemes in small-scale and large-scale scenarios.The main work content is as follows:(1)For WSNs in small-scale application scenarios,a hybrid LEACH clustering routing algorithm based on K-means clustering and adaptive ant colony algorithm is proposed.Aiming at the irrationality of LEACH algorithm to randomly select cluster heads,K-means clustering is used to complete the allocation of initial clusters.Considering that the cluster head plays the role of collecting,processing,and forwarding tasks,it is easily affected by parameters such as communication path quality and node remaining energy.Therefore,based on the cluster head election formula of the LEACH algorithm,the optimal path of the cluster head to the base station,the remaining energy of the node,and the fitness parameters are introduced to construct the election objective function,thereby deriving a new cluster head election threshold function,and then selecting each The best cluster head among the clusters.Among them,the optimal path is searched by the ant colony algorithm.Considering that the path optimization of the ant colony algorithm is likely to cause local optimization due to too fast convergence,redefine the relationship between the initial pheromone concentration and the path selection to achieve an adaptive Effect,and derive a new pheromone concentration update formula.The simulation results show that compared with the traditional method,the proposed method distributes clusters more evenly and effectively balances the network load,while the cluster head performs optimized elections within the cluster,which improves the efficiency of algorithm operation,effectively alleviates the speed of node death and improves network survival life.(2)For WSNs in large-scale application scenarios,considering that the clustering algorithm in the hybrid LEACH routing protocol in small-scale scenarios cannot meet higher performance requirements,a clustering algorithm based on RBF neural network and copula correlation analysis is proposed.Different from the previous scheme,this scheme first conducts cluster head elections.Taking into account the differences in the weights of factors affecting cluster head elections,the fuzzy comprehensive judgment method in mathematical statistics is used to deduce that the remaining energy and distance can be satisfied at the same time.The objective function of the four influencing factors,the data processing ability and the number of nodes in the cluster,and the global fitting ability of the RBF neural network are used to optimize the cluster head for this function.After the cluster head is selected,it will enter the clustering process.The cluster head will invite to join the cluster on the network,measure the amount of information before and after the response node joins the cluster where the cluster head is located,and use the copula function to calculate the correlation between the amount of information,and compare the result with The corresponding threshold is set for comparison.After a judgment,only the nodes that meet the requirements are allowed to become cluster members.Simulation results show that,compared with traditional methods and small-scale improved methods,the proposed method can increase the optimization effect of cluster head selection,increase the processing speed,and effectively reduce network energy consumption.