Research On DDoS Attack Detection Method Based On Convolutional Neural Network

With the rapid development of technologies such as computers,digital communications and networks,it brings us new opportunities and new challenges,but also brings great risks,such as the socialization,liberalization and openness of information networks.Features such as cross-borders are more closely linked around the world,but the lack of legal constraints makes it a place of extra-legality.Today's cyber security field faces three major threats,including worm(Worn)virus,distributed denial of service(DDoS)and spam(Spain).The panic caused by these three has attracted the attention of all countries in the world.Distributed Denial of Service(DDoS)is the most common network attack method.It is the easiest to implement,the most difficult to prevent,and the most difficult to track.It greatly affects the development of the Internet,and an effective service provided by the network service host system.In recent years,deep learning research has made exciting breakthroughs in speech recognition,text processing(such as word segmentation,word segmentation),computer vision and other applications,such as information extraction,machine translation and so on.In view of the extensive application of deep learning in the above,it is applied to DDoS attack detection.The main research contents of this paper are as follows:(1)For BP neural network,there are some shortcomings such as slow learning speed,easy to fall into local minimum value,limited network layer and over-fitting phenomenon during operation,which leads to the degradation of network performance and hinder the improvement of network learning ability.The convolutional neural network is introduced,and the local receptive field,weight sharing and pooling of the convolutional neural network are used to improve the learning ability,expression ability and network performance of the neural network.(2)For the gradient descent method used in convolutional neural networks,it is easy to fall into the local optimum problem because of its fast convergence speed.The particle swarm optimization algorithm is introduced,and the particle swarm optimization algorithm is simple and the ability of searching ability is combined with convolutional neural network.A method of optimizing neural network is proposed to improve the learning ability and network performance of convolutional neural networks.(3)Designed a prototype of forensic system based on convolutional neural network.Based on the data collection and analysis data,the system designs different modules according to different requirements,and completes the functions of the corresponding modules.Convolutional neural network detection based on particle swarm optimization and DDoS detection based on convolutional neural network are applied to the system.