Fault Analysis and Resilience for SpaceCraft Autonomous Protection Enhancement (FARSCAPE)

NASA relies heavily on the computational systems of spacecraft to perform missions in the harsh environment of space. Failures due to faults may interfere with mission-critical events, jeopardizing mission success and potentially resulting in the loss of expensive equipment and scientific data. To ensure mission success, space platforms must be equipped with onboard capabilities to autonomously detect, analyze, and mitigate faults using in-the-loop mechanisms and algorithms. When system-wide faults occur that cannot be mitigated purely in software, the software-based fault management algorithms must detect these and initiate failover to existing hardware-based recovery mechanisms if available on the platform.To address fault management for space platforms, Charles River Analytics is pleased to propose the Fault Analysis and Resilience for SpaceCraft Autonomous Protection Enhancement (FARSCAPE) system, which will automatically monitor, detect, diagnose, and mitigate faults in space platforms to realize greater system autonomy. FARSCAPE adds software-based autonomous fault management capabilities to space systems to (1) monitor system-level, middleware, and application-level mission-critical component execution; (2) apply an ensemble of machine learning (ML) techniques to accurately detect known and unknown faults; (3) compute mitigation plans when faults arise; and (4) execute mitigations to fight through faults and minimize data loss and damage caused by system failures. The proposed FARSCAPE effort is at the intersection of several of our teamrsquo;s longstanding interests, namely fault management and resilient systems design. Our approach is founded on our extensive domain expertise, as well as current and past research at Charles River Analytics in resilient OS design, ML, anomaly detection, system monitoring, and fault resilience.