HIERARCHICAL SECURITY ARCHITECTURE FOR FUTURE NETWORKS

A new mixed-initiative hierarchical framework of humans and agents that is well suited to protecting computational infrastructures. The framework, Cooperative Infrastructure Defense (CID), is designed to rapidly adapt to new cyber attacks via swarming software agents while enabling humans to supervise the system at an appropriate level. We interpose a hierarchy of rational software agents between the swarm and the human supervisors to provide a channel for system guidance and feedback. Using various kinds of rationality actually turns false positives into beneficial forms of positive feedback and improves system performance. CID represents a revolutionary new way of looking at system control for cyber security. Traditionally, humans control the entire system assisted by automated tools and subsystems. In CID, agents share the decision-making power, handling most of the real-time portion autonomously but enabling human involvement at all levels. The human supervisor does not directly control the system rather humans exert supervisory influence sharing the initiative for action with their software agents. CID is designed to be a scalable, dynamic, and robust framework for securing increasingly complex computational infrastructures.M.S. in Electrical Engineering, July 201