Find in Library

With the complexity of network structure, the requirements for network architecture are also increasing, and Software Defined Network (SDN) technology has emerged. SDN technology has successfully simplified network management, but its open programming nature poses a risk of network attacks. In complex network environments, the recognition accuracy of traditional recognition models cannot meet the requirements of accuracy and speed. In view of this, this research proposes an attack identification model based on Convolutional neural network (CNN), hoping to solve the attack identification problems faced in the SDN environment, improve the accuracy of the model, and ensure the security of SDN. An SDN attack recognition model is constructed using the NSL-KDD dataset and the MITLL DARPA dataset, and the CNN is used to utilize it in SDN. In the performance testing experiment of the model, the results show that the proposed model has an accuracy of 98.25% in SDN attack recognition, and its performance is significantly better than traditional CNN models. The accuracy of traditional attack recognition reaches 98.25%, and its performance is superior to the KNN-PSO model. The superiority of the model has been verified, further confirming the application value of the research model in SDN attack recognition.