Please use this identifier to cite or link to this item: https://dspace.iiti.ac.in/handle/123456789/4722
Title: Error masking of transient faults: Exploration of a fault tolerant datapath based on user specified power and delay budget
Authors: Sengupta, Anirban
Keywords: Budget control;Escherichia coli;Faulting;Transients;BFOA;DSE;Equivalent functions;Exploration process;Fault tolerant algorithms;HLS;Multi cycle;Triple modular redundant;High level synthesis
Issue Date: 2014
Publisher: Institute of Electrical and Electronics Engineers Inc.
Citation: Sengupta, A., & Bhadauria, S. (2014). Error masking of transient faults: Exploration of a fault tolerant datapath based on user specified power and delay budget. Paper presented at the Proceedings - 2014 13th International Conference on Information Technology, ICIT 2014, 345-350. doi:10.1109/ICIT.2014.15
Abstract: This paper presents a novel exploration process of an optimal fault tolerant data path based on user specified power and delay budget during high level synthesis (HLS) that is capable of masking error occurred through single and multi cycle transient faults. The exploration framework is driven through bio-mimicking of E. Coli bacterium lifecycle. The major novelties of this approach are as follows: a) novel multi-cycle fault tolerant algorithm, b) novel design space exploration (DSE) approach that combines proposed fault tolerant algorithm along with user specified conflicting power-delay constraint that guides this intractable search problem to reach an optimal solution, c) novel double modular redundant (DMR) system with equivalent circuit scheme that performs the equivalent function of extracting the correct output as conventionally done using the concept of triple modular redundant (TMR) and voter. Results indicated an average improvement in quality of results (QoR) of >24% and reduction in hardware usage of > 57 % were obtained compared to a recent similar approach. © 2014 IEEE.
URI: https://doi.org/10.1109/ICIT.2014.15
https://dspace.iiti.ac.in/handle/123456789/4722
ISBN: 9781479980840
Type of Material: Conference Paper
Appears in Collections:Department of Computer Science and Engineering

Files in This Item:
There are no files associated with this item.


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

Altmetric Badge: