UM MODELO DE ATUALIZAÇÃO AUTOMÁTICA DO MECANISMO DE DETECÇÃO DE ATAQUES DE REDE PARA SISTEMAS DE DETECÇÃO DE INTRUSÃO

Abstract

The obsolescence of the IDS's attack identification mechanisms critically compromises the security level of the networks. This research work presents a proposal of a automatic updating model of the network attack detection mechanism for intrusion detection systems based on a society of intelligent agents. The Security Central Agency, a component of the model, distributes a mini-society of attack detection agents, called SAARA, that uses a neural network trained with data captured from several network traffic sources. A computational implementation of the SAARA model, focusing data driven attack detection, is presented for the multiagent IDS NIDIA.