Triple Modular Redundancy is a widely used fault-tolerance methodology for highly-reliable electronic systems mapped on SRAM-based FPGAs. However, the state-of-the-art TMR techniques are unable to effectively deal with cross-domain errors and increased scrubbing time due to growing size of configuration memory. In order to deal with the aforementioned problems, this work proposes a TMR architecture that exploits the fracturable nature of Look Up Tables for simultaneously mapping of majority-voting and error detection at the granularity of TMR domains. An associated CAD flow is developed for partial reconfiguration of TMR domains incorporating changes to the technology mapping, placement and bitstream generation phases. Our results demonstrate that we can achieve significant reduction in repairing times along with better resilience to cross-domain errors with zero hardware overhead compared to the existing TMR methodologies.
A novel tool-flow for zero-overhead cross-domain error resilient partially reconfigurable X-TMR for SRAM-based FPGAs / Sterpone, Luca; Andres Cardona, Luis; Ullah, Anees; Ferrer, Carles. - In: JOURNAL OF SYSTEMS ARCHITECTURE. - ISSN 1383-7621. - ELETTRONICO. - 81:(2017), pp. 112-120. [10.1016/j.sysarc.2017.10.009]
A novel tool-flow for zero-overhead cross-domain error resilient partially reconfigurable X-TMR for SRAM-based FPGAs
Luca Sterpone;
2017
Abstract
Triple Modular Redundancy is a widely used fault-tolerance methodology for highly-reliable electronic systems mapped on SRAM-based FPGAs. However, the state-of-the-art TMR techniques are unable to effectively deal with cross-domain errors and increased scrubbing time due to growing size of configuration memory. In order to deal with the aforementioned problems, this work proposes a TMR architecture that exploits the fracturable nature of Look Up Tables for simultaneously mapping of majority-voting and error detection at the granularity of TMR domains. An associated CAD flow is developed for partial reconfiguration of TMR domains incorporating changes to the technology mapping, placement and bitstream generation phases. Our results demonstrate that we can achieve significant reduction in repairing times along with better resilience to cross-domain errors with zero hardware overhead compared to the existing TMR methodologies.Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/11583/2693287
Attenzione
Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo