Automating real-time recovery from radiation-induced errors with byte-sized corrections

Radiation-induced errors in high-energy physics detectors threaten data integrity and uptime, posing significant challenges for reliable operation during the future high-luminosity LHC runs. With over 150,000 ASICs to be deployed in the CMS HGCAL front-end, even a very low per-chip error probability results in a non-negligible failure rate. This work presents a firmware-based, byte-level autorecovery mechanism for the CMS HGCAL Phase-2 front-end electronics, capable of detecting transient upset-related errors and restoring chip operation in real time without requiring a full system reset, ensuring data acquisition continues uninterrupted. The system, implemented in FPGA logic and validated using front-end emulation, executes recovery actions within 4.7 µs, corresponding to the measured interval from the data acquisition signal to packet arrival. Overall, the hardware approach is measured to recover approximately 73.5 ms faster than software-based intervention. The results demonstrate robust, autonomous operation in radiation-prone environments while preserving continuous data acquisition.