Research On Botnet Detection Model Based On Support Vector Machine Improved By Optimal Foraging Algorithm

Today,the most important asset for individuals and businesses is information.Botnet is one of the biggest online security threats today.According to the historical botnet attack traffic training classification model,it is very important to predict botnet security threats in the near future in order to prevent botnets and protect the network security.In this paper,the related theories of botnet detection and the commonly used prediction models are studied in detail,and a botnet detection model based on the optimal foraging algorithm is proposed.The main work is as follows:(1)Through the research on the current botnet research dataset and feature selection,we find that most of the botnet datasets nowadays generally contain less samples and more pertinence,which can not fully reflect the actual situation of the botnet.In this paper,the data generated by different malwares in 13 scenarios of CTU-13 are randomly extracted,and a more comprehensive and factual botnet data set is synthesized by using the superposition method.(2)At present traffic-based botnet detection methods,the value of traffic characteristics is not clear,especially under certain data sets and based on specific detection methods.Analyze the characteristics of each flow used in the current study and select one by one according to the strategy of Wrapper methods to extract the most effective subset of features that more closely matches the training method and will yield the best classification accuracy.(3)The advantages and disadvantages of the optimal foraging algorithm are introduced and the parameters C and gamma selection of Gaussian radial basis function support vector machines are optimized.According to the characteristics of dataset,the botnet classification model structure is designed to construct the optimal foraging algorithm to optimize SVM botnet classification prediction model.(4)Through the simulation experiment and comparison with other forecasting models,the optimal foraging algorithm is used to verify the accuracy and performance of the SVM botnet classification model.Simulation results show that the prediction model proposed in this paper is superior to otherprediction models in predicting the botnet's accuracy.