Performance of rule-based system CLIPS on microcontrollers

Abstract Satellite failure detection, isolation, and recovery (FDIR) is critical for mission autonomy, yet its implementation on resource-constrained nanosatellites presents significant challenges. This paper addresses these challenges by presenting a rigorous performance evaluation of a flexible, CLIPS-based FDIR system on a diverse suite of commercial and space-relevant microcontroller architectures, including ARM Cortex-R5F, M7, M4, M0+, and the SPARC V8 (LEON3). Our empirical findings establish the viability of this knowledge-based approach on modern, low-cost hardware, demonstrating that its practical application is governed by a clear set of performance trade-offs. The results reveal a wide performance spectrum, with rule execution times for nominal monitoring tasks ranging from approximately 60 $$\upmu$$ μ s on a high-performance ARM Cortex-M7 to over 4000 $$\upmu$$ μ s on a simulated LEON3. We identify the data interface for fact creation and assertion as a primary performance bottleneck, often consuming more time than the core reasoning process. This research provides tangible data to guide satellite system architects in selecting appropriate hardware, confirming that expressive, rule-based FDIR is a feasible and resilient solution for the next generation of small spacecraft