Algoritmos Imunoinspirados aplicados em segurança computacional: utilização de algoritmos inspirados no sistema imune para detecção de intrusos em redes de computadores

Abstract

THE evolution of Internet and the continuous interconnection of computersand other digital devices has opened many opportunities. On the otherhand, this process has brought a lot of problems and pitfalls. One of theseproblems refers to the increasing number of computers being invaded by intruders.The work presented in this text approaches this problem by adoptingstrategies and algorithms inspired by concepts and processes of the inate andadaptive immune system. We have proposed two algorithms, IA-AIS and DTAIS,and they were evaluated in datasets simulating a computer in normaluse and under attack. IA-AIS was also evaluated with KDD99, a publiclyavailable dataset used in the literature to comparative analysis. The obtainedresults were very good, indicating the adequacy of the use of more completemetaphors of the immune system in the intrusion detection problem. Moreover,these results showed the algorithms as very promising tools in anomalydetection problems, indicating their great potential application in productionsystems.