Research On Intrusion Detection Of Wireless Sensor Networks Based On Hierarchical Structure

With the continuous maturity of the ecological environment of the Internet of Things,the development speed of the Wireless Sensor Network(WSN)is gradually accelerating.Compared with other traditional networks,WSN is more vulnerable to various intrusions because sensor nodes in WSN have limitations such as limited energy,weak wireless communication capabilities,and weak computing capabilities,and are usually deployed in very harsh and complex geographical environments.Behavior damage,thus causing serious network security problems.As one of the most commonly used security protection measures in the field of network security,intrusion detection technology can effectively guarantee the security of WSN to a certain extent.Considering the differences between WSN and other networks,the existing intrusion detection algorithms cannot be directly applied to WSN.Therefore,this thesis proposes an intrusion detection scheme suitable for WSN based on the hierarchical network structure based on the above-mentioned limitations in WSN.The main contents are as follows:1.Aiming at the problem of limited node energy in WSN,this thesis proposes an intrusion detection scheme based on hierarchical network structure.The core of the scheme includes cluster head nodes and base station nodes.According to the difference in the amount of node resources,the intrusion detection work can be completed collaboratively to avoid severe energy consumption of local nodes,thereby improving the life cycle of nodes and networks.2.Aiming at the problem of limited computing power of cluster head nodes and high dimensionality of WSN data,in this thesis,combined with machine learning algorithms,a WSN intrusion detection model based on Principal Component Analysis(PCA)and Support Vector Machine(SVM)is proposed at the cluster head node.By comparing the detection results of different data dimensions with the SVM classification algorithm,considering the detection accuracy and time complexity,the best data dimension is selected.The experimental results based on the WSN-DS data set show that the detection accuracy of the model reaches97.08%,which is only 0.01% lower than that of the SVM algorithm,but it greatly reduces the time complexity of the model.3.Based on the detection results of cluster head nodes,in order to improve the multiclassification effect of WSN intrusion detection,combined with the deep learning algorithm,this thesis proposes a WSN intrusion detection model based on Convolutional Neural Networks(CNN)and Gated Recurrent Unit(GRU)at the base station node.The model can fully learn the spatial and temporal characteristics of the data to improve the detection accuracy and multi-classification effect.The experimental results based on the WSN-DS data set show that the detection accuracy of the model reaches 99.57%,and it is also superior to the existing WSN intrusion detection algorithm in terms of time complexity and multiclassification effect.