Application Of Ant Colony Algorithm In Network Intrusion Feature Selection

With the rapid development of Internet technology,network intrusion is more frequent and more complex.However,the firewall,which is widely considered to be preventing intrusion,have a number of shortcomings and cannot deal the growing network security crisis.Real-time detection system has become an effective complement to the firewall.In general,feature informations are always over-extracted to improve the recognition rate of the intrusion detection system classifier.but it leads to a dramatic increase in the dimension of the feature space,and makes the feature space redundant.Therefore,there is an urgent need for a program,which can reduce the feature space dimension and improve the classification accuracy of the program,to further improve the performance of intrusion detection.In this thesis,ant colony optimization(ACO)and genetic algorithm(GA)are used to improve the feature selection method of network intrusion,which is applied in the field of intrusion detection to improve the performance of detection.The main contents of this thesis include:To overcome the problem of stagnation and local optimal,an strategy of dynamically adjusting parameters is presented.On the one hand,in order to avoid that the denominator value is too small to be treated in a certain precision range when there is no ants passing through some paths,the thesis introduce the same order of magnitude to expand the pheromone concentration value and the heuristic function value in the selection probability calculation formula.On the other hand,the pheromone volatilization coefficient is changed in real time according to the pheromone concentration value in the search process,so that the higher the pheromone concentration on the path,the faster the volatilization,the lower the concentration and the slower the volatilization.Ants can search in a larger range by balancing the pheromone concentration of each path.To overcome the problem of ant colony optimization shorting of the pheromone in the initial of the intrusion detection feature selection,the thesis propose a strategy of combining the improved ant colony optimization with genetic algorithm.At the beginig of the search,the genetic algorithm is used to quickly search the feature set.According to the screening results,the ant colony optimization is targeted to set a certain pheromone initial value,so as to improve the speed of ant colony convergence.The experimental results show that the proposed strategy can effectively eliminate the feature redundancy in feature selection,improve the correctness of detection,reduce the false alarm rate and false negative rate,and achieve the balance between time efficiency and detection efficiency.