US eyes artificial intelligence-based cyber defenses for tactical networks

U.S. Army researchers are surveying the defense industry to find companies able to develop autonomous cyber defensesfor tactical networks and communications that capitalize on artificial intelligence and machine learning.




Officials of the Army Contracting Command at Aberdeen Proving Ground, Md., issued a request for information (W56KGU-19-R-AUTOCYBER) on Monday for the Autonomous Cyber project.

Researchers are looking for cyber technology to secure automated network decisions and defend against adaptive autonomous cyber attackers at machine speed.

The Army Contracting Command is conducting this industry survey on behalf of the Army Communications-Electronics Research, Development and Engineering Center Space and Terrestrial Communications Directorate (S&TCD).

Overall, S&TCD envisions a combination of several artificial intelligence and machine learning products that deliver autonomous cyber defense capabilities. Specifically, researchers are looking for cyber and trusted computing enabling technologies for:

-- autonomous detection and mitigation of known cyber vulnerabilities;

-- ways to autonomously identify and correct misconfigurations in networks and hosts;

-- ways to autonomously detect known and previously unknown malware samples;

-- tools and methodologies for red team autonomous decision making engines;

-- ways to improve robustness of autonomous decision engines to manipulate attackers;

-- machine learning-based cyber agents tailored to specific tactical networks, data flows, and message sets that can detect and deduce the intent of an attack;[Native Advertisement]

-- an interface that capitalizes on human in the loop feedback to autonomous decision engines to improve the performance and efficiency of human-machine teams; and

-- new ways to correlate cyber response recommendations and generate a course of action based on available cyber tools and cyber events.


White papers should describe how to meet the above goals, and include any available independent performance modeling and test results, as well as hardware specifications, operating systems, and architectures.