A fault-tolerant system-on-programmable-chip based on domain-partition and blind reconfiguration | IEEE Conference Publication | IEEE Xplore

A fault-tolerant system-on-programmable-chip based on domain-partition and blind reconfiguration


Abstract:

Field programmable gate arrays (FPGAs) are widely used in building Systems-on-Programmable-Chips (SOPCs) since they contain plenty of reconfigurable heterogeneous resourc...Show More

Abstract:

Field programmable gate arrays (FPGAs) are widely used in building Systems-on-Programmable-Chips (SOPCs) since they contain plenty of reconfigurable heterogeneous resources providing the facility to implement various intellectual property cores. However, with the shrinking device feature size and the increasing die area, nowadays FPGAs can be deeply affected by the errors induced by electromigration and radiation, which results in challenges of building reliable SOPCs. In this paper, a SOPC implementing a smart 1553B bus node is presented to investigate the challenges and illustrate a feasible approach for building a complex system aimed at high reliability and low recovery latency on a commercial FPGA. First, a general reliability model, the DomainPartition (DP) model, is introduced to formulate the SOPCs which contain multiple alternative configurations proving the fault recovery capability. The assignment of the alternative configurations for maximizing the reliability is then determined according to a first-order optimal solution under the DP framework. Finally, the blind reconfiguration technique is used to reduce the recovery latency. The experiments based on a Monte Carlo simulation approach are carried out to evaluate the reliability and the latency. The obtained results show that higher reliability is attainable with less overhead than the generic triple-modular redundancy method.
Date of Conference: 15-18 June 2010
Date Added to IEEE Xplore: 12 August 2010
ISBN Information:
Conference Location: Anaheim, CA, USA

References

References is not available for this document.