Tactical Deployment and Management of Adaptive Agents
Principal Investigator: G Fink
Technical Advisor: G Fink, Adaptive Systems Focus Area
Purpose of research
- Enables member organizations and agencies in critical infrastructures to mutually defend their computational assets
- Supports the awareness, prevention, and response functions of DHS
Couple human intelligence with a hierarchy of rational software agents to cooperatively protect an infrastructure of independent organizations and agencies
Makes humans an intrinsic part of the solution, without requiring their direct control. Enables diverse organizations to cooperate in an adaptive cyber defense.
Today's critical infrastructures are supported by vast, interwoven networks of machines in a multi-organizational cyber infrastructure. To succeed against concerted attacks by our adversaries, these infrastructures must cooperatively self-defend as an organic whole.
Humans retain ultimate responsibility for the actions of their systems, but they are unable to act at the speed of automated attacks. Humans will supervise autonomic defense networks of software agents by providing high-level guidance. Humans will be required to authorize only decisions that are beyond the abilities of their agents to perform automatically.
Top-level, rational software agents translate the human-provided guidance into executable, unambiguous network policy for the organizational units they are charged with. These agents are logic-based, and can explain their actions to the human supervisor if required. The top-level agents provide a window for the human supervisor into the activity of the autonomic defense.
At the level of each machine, another kind of agent interprets the policy, detects problems, and makes repairs as needed. These agents are represented by the large shaded circles under each managed machine. In addition to these, there are swarming mobile agents (represented in the drawing by the small, dark circles surrounding various machines) that are programmed to detect certain types of problems. The swarm-based intelligence of these mobile agents uses adaptations of social insect behaviors to solve the highly constrained problems of detecting previously unknown threats. These swarming agents share knowledge of potential problems across machines within the organizational unit, but do not go outside its boundaries.
TDMAA, or the Cooperative Infrastructure Defense unifies human, logic-based, and swarm-based reasoning to provide autonomic defense with human influence.
Learn more about Tactical Deployment and Management of Adaptive Agents