Initial Development of Embedded Low-Power Parallel Processing for On-Orbit Spacecraft Anomaly Management

Maintaining the health of a spacecraft is a critical element of satellite operations, typically driving operations costs and often requiring standing armies of highly trained operations staff. An important aspect of this work is anomaly management (AM), which is the detection, diagnosis and resolution of anomalous conditions. Research in this area ranges from the development of robust reasoning techniques to the design of highly-performance flight processors able to implement these techniques on-orbit. Our recent work in this area focuses on the composition of advanced model-based reasoning (MBR) algorithms for AM that can be efficiently executed on a new generation of lowpower, low-cost multi-core embedded processors. These processors provide a parallel-processing capability that can potentially revolutionize the performance of highly accurate but deliberative and computationally expensive MBR algorithms while still being suitable for space vehicle applications. In this paper, we will describe our latest theoretical and algorithmic contributions to MBR-based AM. We will discuss how our successfully demonstrated algorithms are being recast for parallel processing, and how these newly formed algorithms are being prototyped using new low-power multi-core embedded processors. Finally, we will discuss testbeds for this technology ranging from ground engineering units to simple student-based flight experiments