Publication Details
Methodology for Reliability Analysis of FPGA-based Fault Tolerant Systems
Kaštil Jan, Ing., Ph.D.
Kotásek Zdeněk, doc. Ing., CSc.
reliability, dependability, Markov model, fault tolerant system, FPGA, reconfiguration, soft-error, duplex architecture
In this paper, a methodology for reliability and availability analysis of fault tolerant architectures based on partial reconfiguration implemented into FPGA is presented. First, the introduction into the system reliability, Markov models and dependability parametres is given. Then, the priciples of methodology for generation of Markov model for duplex architecture with checkers are shown together with the evaluation of failure rate and repair rate of this architecture. Finally, it is demonstrated how the reliability and availability parameters are derived from this model for different configurations of architecture and faulty modules. The reliability analysis results are shown in the graphs.
In this paper, a methodology for reliability and availability analysis of fault tolerant architectures based on partial reconfiguration implemented into FPGA is presented. First, the introduction into the system reliability, Markov models and dependability parametres is given. Then, the priciples of methodology for generation of Markov model for duplex architecture with checkers are shown together with the evaluation of failure rate and repair rate of this architecture. Finally, it is demonstrated how the reliability and availability parameters are derived from this model for different configurations of architecture and faulty modules. The reliability analysis results are shown in the graphs.
@inproceedings{BUT96988,
author="Martin {Straka} and Jan {Kaštil} and Zdeněk {Kotásek}",
title="Methodology for Reliability Analysis of FPGA-based Fault Tolerant Systems",
booktitle="CSE'2012 International Scientific Conference on Computer Science and Engineering",
year="2012",
pages="146--153",
publisher="The University of Technology Košice",
address="Košice",
isbn="978-80-8143-049-7"
}