Soft computing tools to detect and characterize anomalous network behavior

Abstract

As part of a funded effort in network security, soft computing is used to create a system capable of detecting and characterizing anomalous network behavior. Soft computing is composed of evolutionary computational tools combined with artificial neural networks and/or fuzzy logic. Actual network traffic is used as opposed to simulated data. The benefits here are that the attacks are real and the use of the tool is in a "production" environment as opposed to being a product of purely simulated theoretical speculation.