Abstract
Single Event Upsets (SEUs) in Commercial Off-The-Shelf (COTS) Software-Defined Radios (SDRs) are frequent in a erospace applications, especially in GEO (Geostationary Orbit) orbit during severe solar activity, and can lead to unexpected register corruption and communication failures. This work presents a purely software-based Fault Detection, Isolation, and Recovery (FDIR) framework tailored for the AD936x RF agile transceiver, requiring no hardware modifications. The proposed method classifies all device registers into four impact categories and applies dedicated scrubbing strategies-standard refresh, masked refresh, procedural refresh, and forced refresh-combined with real-time register health monitoring and adaptive recovery actions. Fault injection experiments comprising 10,000 diverse test cases achieved 100% fault coverage for the tested scenarios, with an average recovery time of 0.75 s for typical SEUs and a guaranteed worst-case recovery under 4.4 s for critical failures, while maintaining a CPU load below 1.3%. The approach ensures continuous SDR operation under SEU events and offers a scalable, lightweight, and cost-effective reliability enhancement for CubeSats and other resource-constrained aerospace platforms.