Fault tolerant Field Programmable Neural Networks

Krcma, Martin; Kotásek, Zdenĕk; Kastil, Jan

doi:10.1109/norchip.2015.7364381

Cited by 7 publications

(1 citation statement)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Typical examples include load balancing and thermal management as well as modular standby sub-components. Finally, the fault-tolerant properties of neural networks have been investigated by incorporating TMR principles into the network [100]. A summary of key design features expected of reconfigurable computer systems can be found in Table XII.…”

Section: B Active Methodsmentioning

confidence: 99%

Zero-maintenance of electronic systems: Perspectives, challenges, and opportunities

McWilliam

Khan

Farnsworth

et al. 2018

Microelectronics Reliability

View full text Add to dashboard Cite

Self-engineering systems that are capable of repairing themselves in-situ without the need for human decision (or intervention) could be used to achieve zero-maintenance. This philosophy is synonymous to the way in which the human body heals and repairs itself up to a point. This article synthesises issues related to an emerging area of self-healing technologies that links software and hardware mitigations strategies. Efforts are concentrated on built-in detection, masking and active mitigation that comprises self-recovery or self-repair capability, and has a focus on system resilience and recovering from fault events. Design techniques are critically reviewed to clarify the role of fault coverage, resource allocation and fault awareness, set in the context of existing and emerging printable/nanoscale manufacturing processes. The analysis presents new opportunities to form a view on the research required for a successful integration of zero-maintenance. Finally, the potential cost benefits and future trends are enumerated.

show abstract

Section: B Active Methodsmentioning

confidence: 99%