Modeling and generation of detectors in artificial immune based network intrusion detection systems

Abstract

In this paper, a novel model is built for detectors, used in network intrusion detection systems. The model works with a newly developed energy function, which is used in calculating the energies of detectors. An algorithm based on Genetic Algorithms, is developed for the generation of proposed detectors. This algorithm plus the model are tested over both synthetic and real data. Testing of attacks is based on the attacks present in the headers of the TCP/IP packets. Some real attacks have been simulated like WinNuke and Ping of death. For both the real and the synthetic cases, the results are illustratedfor a case ofjust less than 2% and 1% of false negatives.