A First Step Towards a General-Purpose Distributed Cyberdefense System

Most part of the activity in cybersecurity consists of perceiving the state of a distributed environment composed of a network of interconnected computers or devices, taking care of every individual perception of a device but also observing the whole picture of the network. When needed, there is a reaction by executing actions aimed at producing some preventive, defensive or restoring effect in that environment in order to keep it in a safe state. This is, nearly by definition, an ideal area of application for multi-agent systems. This paper presents the first steps given to build a general-purpose multi-agent system for the cyberdefense of a network of computers. It also shows a preliminary experiment to measure the burden of implementing this distributed defense mechanism in terms of CPU and memory overload.