Research On Network Evasion Detection Based On Deep Learning Technology

In recent years,with the rapid development of computer technology and the high popularity of the Internet,the network has become an indispensable carrier for information interaction.However,the network environment is becoming more and more complex,and security issues are becoming more and more prominent.If a cyber attack is successfully invaded,it will not only cause data loss,but also cause major security problems,causing significant loss of life and property to society and the country.Enterprises are introducing Network Intrusion Detection/Prevention Systems to monitor network flows,by which they want to see whether the anomalies and misuses are occurring.However,evasion is able to disguise attacks to avoid detection and blocking by the network security system.Evasions can be a way to confuse network intrusion detection systems by masking data traffic and also be applied to normal traffic,as well as to attack.They are considered successful as long as the delivery mechanism succeeds in gaining access to victim computers while the security device fails to detect or respond to the attack.Besides that,there are not effective ways to detect network evasions when faced with a large volume of network flows,existing methods have various deficiencies.Therefore,how to detect network evasion behavior is the focus of current academic research.Network evasion is a way to confuse network intrusion detection systems by masking data traffic.Network evasion detection aims to distinguish whether network traffic from the link layer poses a threat to the network.At present,the traditional network evasion detection method does not extract the characteristics of network traffic,and the detection accuracy is relatively low.This article will discuss the atomic evasion that constitutes advanced evasion technology.Firstly,it will expound the background and significance of the topic selection,the current research status and development of network escape technology at home and abroad,and the harm it brings.Then,this paper will comprehensively analyze the implementation of evasion techniques and introduce the atomic technology of network evasion from the TCP/IP layer.Secondly,a feature extraction algorithm suitable for escaping behavior detection is proposed and implemented with the method of generating network evasion data stream samples is designed and implemented.Finally,A new network evasion detection algorithm based on deep recurrent neural networks is proposed to detect eight atomic evade behaviors.A large number of experimental results on the test set show that the proposed algorithm has higher recognition accuracy of the network escape behavior,and provides an idea for the Internet and the smart network internal network intrusion/defense detection system.