Development of algorithm for intrusion detection system based on artificial immune systems

The rapid development of data transfer systems, the advent of the latest access technologies and advanced mobile devices require adequate growth of computer and network security. Traditional algorithms lose their effectiveness with the increasing complexity of protected systems, so research into bio-inspired methods to increase the efficiency of intrusion detection systems is relevant. The use of methods based on artificial immune systems involves finding a compromise between the accuracy and speed of intrusion detection. This article is devoted to the development of an algorithm containing several procedures: filtering the input set of data, selecting parameters of network activity using the theory of approximate sets and genetic programming, generating detectors using negative selection. Simulation experiments consisted of detecting anomalies using the approximate set method and using a genetic algorithm and comparing network activity data with rows contained in the detector. The results show that the method of approximate sets has the worst result in detecting attacks, but the best in screening out false attacks, and the linear genetic picture is the opposite, which indicates the need for further research.