Perspecta Labs wins autonomous defensive cyber operations contract
Chantilly, VA-based Perspecta Inc. announced on April 22 that its innovative applied research arm, Perspecta Labs, was awarded the Autonomous Defensive Cyber Operations program from the U.S. Army Combat Capabilities Development Command (CCDC), Command, Control, Computers, Communications, Cyber, Intelligence, Surveillance and Reconnaissance Center (C5ISR), Space and Terrestrial Communications Directorate. The program, which represents new work for the company, has a five-year period of performance, including options and total potential value of $14.5 million.
The objective of the program is to develop a suite of cyber capabilities for the U.S. Army that are autonomous, secure, easy to deploy and configure, adaptive and resilient to adversarial deception. On the program, Perspecta Labs will research, design, develop, demonstrate and deliver a machine learning for defensive cyber operations solution for training, rapidly deploying and retraining containerized cyber sensors that detect both known and unknown vulnerabilities, attacks and malware. These highly autonomous capabilities are lighter weight and much quicker to respond than traditional approaches that require human intervention which is critical in the dynamic and resource-constrained tactical environment.
“Perspecta Labs will leverage its extensive expertise in machine learning, cybersecurity and tactical networking to provide innovative and effective autonomous defensive cyber operations to the Army,” said Petros Mouchtaris, Ph.D., president of Perspecta Labs. “We are honored to have been selected to develop a solution that meets the Army’s tactical needs for autonomy, security, ease of use, adaptability, efficiency and robustness.”
Perspecta Labs’ solution will utilize the company’s novel machine learning paradigm called LUPI (Learning Using Privileged Information) to create highly accurate cyber sensors in a heterogeneous environment. These cyber sensors are resilient to adversarial machine learning attacks and can be efficiently and dynamically retrained to take into account the large variations present in the tactical environment. The solution’s architecture harnesses these cyber sensors to detect diverse attacks and incorporates a cognitive agent for automatically generating courses of action for the cyber defender.