Enhancing the Reliability of AD936x-Based SDRs for Aerospace Applications via Active Register Scrubbing and...

Area of Science:

• Aerospace Engineering
• Computer Engineering
• Reliability Engineering

Background:

• Single Event Upsets (SEUs) frequently impact Commercial Off-The-Shelf (COTS) Software-Defined Radios (SDRs) in aerospace, particularly in Geostationary Orbit (GEO) during solar events.
• These upsets can corrupt registers and cause communication failures, posing a significant risk to mission-critical systems.

Purpose of the Study:

• To develop and evaluate a purely software-based Fault Detection, Isolation, and Recovery (FDIR) framework for the AD936x RF transceiver.
• To enhance the reliability of SDRs against SEUs without requiring any hardware modifications.

Main Methods:

• Classifying device registers into four impact categories.
• Implementing dedicated scrubbing strategies: standard, masked, procedural, and forced refresh.
• Integrating real-time register health monitoring and adaptive recovery actions.
• Conducting fault injection experiments with 10,000 test cases.

Main Results:

• Achieved 100% fault coverage for tested SEU scenarios.
• Demonstrated an average SEU recovery time of 0.75 seconds.
• Guaranteed worst-case recovery under 4.4 seconds for critical failures.
• Maintained CPU load below 1.3%.

Conclusions:

• The software-based FDIR framework ensures continuous SDR operation during SEU events.
• The approach provides a scalable, lightweight, and cost-effective reliability enhancement.
• This solution is ideal for resource-constrained platforms like CubeSats in aerospace applications.