Traffic Monitoring And Analysis Of Campus Network Based On Machine Learning

At present,Internet technology is developing rapidly,and the construction of college campus network is also keeping up with the pace of development of the times.However,the rapid development of the network has also brought about various security problems.How to accurately and quickly identify the abnormal traffic in the network traffic has become the focus of the research direction of network traffic monitoring and analysis.With the continuous increase in the number of campus network users,the explosive growth of campus network traffic has greatly increased the difficulty of monitoring and analyzing campus network traffic.The existing campus network traffic monitoring system is difficult to deal with the increasingly complex various traffic attack methods.According to the needs of campus network traffic monitoring in the new era,this paper designs a set of campus network traffic monitoring and analysis system based on machine learning,which integrates campus network traffic monitoring and detection of abnormal traffic.The main work of this paper is as follows:Comprehensive analysis of the pros and cons of various data collection technologies,this article finally selects NetFlow(Cisco Traffic Collection Protocol)technology based on the design of measurement analysis module.In this paper,an optimized combination of Isolation Forest algorithm and optimized K-means(K-means clustering)algorithm is designed to design a set of algorithms.This paper designs a combined algorithm that first divides the network traffic into normal and abnormal through the isolated forest algorithm,and then optimizes the K-means clustering algorithm to fine-grain the abnormal network traffic classification.Then use the real data set KDD CUP99 as the simulated traffic to conduct experiments to analyze the improvement of the performance of the network traffic monitoring system by the machine learning combination algorithm.It is preliminarily verified that the machine learning combination algorithm can not only achieve a higher detection rate with a short time cost,but also make up for the insufficiency of the isolated forest algorithm in detecting abnormal types.The campus network traffic monitoring and analysis system designed in this paper is used in the actual campus network.The operating results show that the overall performance of the campus network traffic monitoring system designed in this paper is improved by 11.2% compared with the original system,which proves that the performance of the system designed in this paper has been improved.The experimental results prove that when using the real data set KDD CUP99 as the simulation traffic input,the average detection rate is increased by 13%,and the average false alarm rate is decreased by 0.36%.During the one-month trial operation of the system,compared with the original system,the detection performance of the system designed in this paper increased by 11.2%.