Consensus-Based Evolvable Hardware for Sustainable Fault-Handling

An autonomous self-recovery approach for SRAM-based Field Programmable Gate Arrays (FPGAs) is developed using Consensus-Based Evaluation (CBE). Under the proposed CBE technique, a diverse population of functionally identical (same input-output behavior), yet physically distinct (alternative design or place-and-route realization) FPGA configurations is generated at design time. At run-time, these alternative configurations are ranked using a fitness function which evaluates an individual’s discrepant behavior relative to the consensus formed by the population. Hence, any physical resource exhibiting an operationally-significant fault automatically decreases the fitness of those configurations which use it. Superior individuals are preferred for selection, so the fault becomes occluded from subsequent FPGA operations. Meanwhile, offspring formed through crossover and mutation of faulty and viable configurations are selected at a controlled re-introduction rate. This enables evolution of a customized fault-specific regeneration, realized directly as new configurations within the FPGA’s normal online data throughput operations. Hence, the Detection, Isolation, Diagnosis, and Recovery phases are integrated into a cohesive evolutionary computing approach emphasizing graceful degradation and device availability. Refurbishments are evolved in-situ, in real-time, without additional test vectors, while allowing the FPGA device to remain partially online. Index Terms Autonomous Fault Handling, Reconfigurable FPGA Devices, Online Evaluation and Refurbishment, Outlier Identification, Genetic Operators for Regeneration.