Design And Implementation Of Network Intrusion Detection System Based On Deep Residual Convolution

The outline of the 14th five year plan proposes to increase investment in big data,industrial intelligence,blockchain and other new infrastructure,and the degree of informatization will be further deepened.With the development of network information technology,however,some people will use technology to bring great threat to people's privacy and property security.Network information security is facing new and complicated challenges.As an important part of network security,intrusion detection system provides timely protection in the face of abnormal intrusion,user misoperation and internal attack,so the research of intrusion detection system has become the key research direction of relevant practitioners.In recent years,with deep learning in speech recognition,image processing,text translation and other fields have achieved good results,various industries are more closely connected with it.At the same time,it also brings new development opportunities for intrusion detection.Based on the powerful feature extraction function of deep learning,this paper studies intrusion detection.The main work of this paper is summarized as follows.(1)According to the temporal and spatial characteristics of traffic data,design and implement the intrusion detection model based on convolutional neural network and the intrusion detection model based on long and short memory neural network respectively.Through comparative experimental verification,it is determined to use convolutional neural network as the basic network for the next step of transformation and optimization.(2)Aiming at the "gradient dispersion/explosion" phenomenon of the deep network model,this paper first builds two residual convolution units based on the residual convolution and Inception structure.The former is used to extract the shallow features of the data,and the latter is used to extract the deep features of the data.Secondly,the Softpool method is used for pooling operation to ensure more data characteristic information.Finally,based on the above design and implement a deep residual convolutional neural network model.At the same time,the experimental verification on the data set shows that the accuracy and detection rate of this model are higher than traditional machine learning algorithms,convolutional neural networks,long and short memory neural networks,and multi-scale convolutional neural network intrusion detection models.(3)Based on the constructed deep residual convolutional neural network intrusion detection model,this paper develops the prototype of the intrusion detection system and performs functional tests on it.The test results show that the prototype of the intrusion detection system can achieve various requirements of intrusion detection.