Stouch, D., Gorman, J., and Ruttenberg, B.
Air Force Research Laboratory Space Situational Awareness Conference, Kihei, HI (September 2017)
Satellites are continually at risk of collision and subject to malicious activity. We must understand the nature (i.e., severity, location, source, timing) of these threats to protect our space assets and respond effectively. A rapid, comprehensive understanding of threat indications and warnings (I&W) is essential to provide as much lead time—with as many response options—as possible. Turning existing data sources that support space situational awareness (SSA) into actionable intelligence to support offensive and defensive space control operations in response to adversary tactics remains a challenge. To overcome this challenge, Charles River Analytics will be developing an analytic tool for Probabilistic Inference for Course of action (COA) Analysis in Space Situational Awareness (PICASSA). Selected for funding under the DARPA Hallmark program, PICASSA will support I&W of potential threats and improve the accuracy of threat indicators. These indicators will be encoded in a probabilistic relational model (PRM) to help shape space defense plans by defining appropriate threshold levels at which to escalate notifications of potential adversarial activity. The analytic tool will produce a probability distribution over possible outcomes (e.g., HVAs being targeted, proximity attacks being used, adversaries behind the attack) to identify the ‘most significant’ threats, which may not be the ‘most likely’ threats. Providing a probability distribution of threats instead of a single most likely threat prevents tunnel vision when reasoning about uncertain situations, provides resilience to data outages, enables forensic assessment of historical events to support attribution, and helps to justify decisions. PICASSA will include a second PRM to evaluate candidate offensive space control (OSC) COAs that assesses their performance to improve successful outcomes by identifying potential limiting factors and specific COA aspects that may lead to detection (and attribution), and recommending adjustments to improve overall COA performance.
For More Information
(Please include your name, address, organization, and the paper reference. Requests without this information will not be honored.)