| With the increasing integration of the Internet and life,the network topology and connected devices are becoming more and more complex,and the importance of network security is becoming more and more prominent.Network detection technology plays an important role in maintaining network security.Its main job is to continuously detect the current network status,detect abnormal behavior status in the network through detection,and promptly alert network management personnel.The timeliness and accuracy of the network detection system are related to the availability and reliability of the current network.This paper first studies the application of deep learning technology in the field of network intrusion detection,with the help of deep learning algorithms to automatically extract the features of intrusion data,to avoid manual filtering of features,and proposes an intrusion detection method based on improved convolutional neural network.The method has mainly been improved in two aspects: On the one hand,the multi-level convolution and multi-level feature fusion of the Inception module are used to enhance the extraction of intrusive data features,and a sparse structure is built in the feature dimension to replace the single feature extraction method of the convolution layer;On the other hand,the optimization of the pooling layer,the establishment of a parallel dimensionality reduction structure to replace the pooling layer dimensionality reduction operation,to avoid the characteristic bottleneck problem that the pooling layer dimensionality reduction may bring.In this paper,the NSL-KDD data set is used to train and evaluate the intrusion detection model of the improved convolutional neural network.In this paper,the NSL-KDD data set is used to train and evaluate the intrusion detection model of the improved convolutional neural network.The simulation results show that the accuracy of the detection method of the improved convolutional neural network and the speed of model convergence are superior to the detection method of the convolutional neural network.The detection method proposed in this paper is an effective and reliable detection method suitable for large-scale intrusion data processing.Secondly,this paper studies network traffic anomaly detection methods,and uses time series autoregressive models to predict network traffic changes.Aiming at the problem of static threshold in the residual ratio detection method,this paper proposes a detection algorithm based on reputation value management,which converts the judgment result of the static threshold into positive and negative feedback,and calculates the cumulative value of the reputation value of continuous observation points to judge network traffic Abnormal behavior.This paper uses OPNET tools to simulate a small network.The simulation results show that the anomaly detection algorithm proposed in this paper has higher accuracy,reduces the false alarm rate,weakens the impact of a single observation point on anomaly detection,and considers the common changes of multiple consecutive observation points Reflects the characteristics of network traffic changes. |
