Fault tolerance issues in nanoelectronics
The astonishing success story of microelectronics cannot go on indefinitely. In fact, once devices reach the few-atom scale (nanoelectronics), transient quantum effects are expected to impair their behaviour. Fault tolerant techniques will then be required. The aim of this thesis is to investigate t...
Main Author: | |
---|---|
Published: |
University College London (University of London)
2008
|
Subjects: | |
Online Access: | http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.564522 |
id |
ndltd-bl.uk-oai-ethos.bl.uk-564522 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-bl.uk-oai-ethos.bl.uk-5645222015-12-03T03:28:02ZFault tolerance issues in nanoelectronicsSpagocci, S.2008The astonishing success story of microelectronics cannot go on indefinitely. In fact, once devices reach the few-atom scale (nanoelectronics), transient quantum effects are expected to impair their behaviour. Fault tolerant techniques will then be required. The aim of this thesis is to investigate the problem of transient errors in nanoelectronic devices. Transient error rates for a selection of nanoelectronic gates, based upon quantum cellular automata and single electron devices, in which the electrostatic interaction between electrons is used to create Boolean circuits, are estimated. On the bases of such results, various fault tolerant solutions are proposed, for both logic and memory nanochips. As for logic chips, traditional techniques are found to be unsuitable. A new technique, in which the voting approach of triple modular redundancy (TMR) is extended by cascading TMR units composed of nanogate clusters, is proposed and generalised to other voting approaches. For memory chips, an error correcting code approach is found to be suitable. Various codes are considered and a lookup table approach is proposed for encoding and decoding. We are then able to give estimations for the redundancy level to be provided on nanochips, so as to make their mean time between failures acceptable. It is found that, for logic chips, space redundancies up to a few tens are required, if mean times between failures have to be of the order of a few years. Space redundancy can also be traded for time redundancy. As for memory chips, mean times between failures of the order of a few years are found to imply both space and time redundancies of the order of ten.500University College London (University of London)http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.564522http://discovery.ucl.ac.uk/14227/Electronic Thesis or Dissertation |
collection |
NDLTD |
sources |
NDLTD |
topic |
500 |
spellingShingle |
500 Spagocci, S. Fault tolerance issues in nanoelectronics |
description |
The astonishing success story of microelectronics cannot go on indefinitely. In fact, once devices reach the few-atom scale (nanoelectronics), transient quantum effects are expected to impair their behaviour. Fault tolerant techniques will then be required. The aim of this thesis is to investigate the problem of transient errors in nanoelectronic devices. Transient error rates for a selection of nanoelectronic gates, based upon quantum cellular automata and single electron devices, in which the electrostatic interaction between electrons is used to create Boolean circuits, are estimated. On the bases of such results, various fault tolerant solutions are proposed, for both logic and memory nanochips. As for logic chips, traditional techniques are found to be unsuitable. A new technique, in which the voting approach of triple modular redundancy (TMR) is extended by cascading TMR units composed of nanogate clusters, is proposed and generalised to other voting approaches. For memory chips, an error correcting code approach is found to be suitable. Various codes are considered and a lookup table approach is proposed for encoding and decoding. We are then able to give estimations for the redundancy level to be provided on nanochips, so as to make their mean time between failures acceptable. It is found that, for logic chips, space redundancies up to a few tens are required, if mean times between failures have to be of the order of a few years. Space redundancy can also be traded for time redundancy. As for memory chips, mean times between failures of the order of a few years are found to imply both space and time redundancies of the order of ten. |
author |
Spagocci, S. |
author_facet |
Spagocci, S. |
author_sort |
Spagocci, S. |
title |
Fault tolerance issues in nanoelectronics |
title_short |
Fault tolerance issues in nanoelectronics |
title_full |
Fault tolerance issues in nanoelectronics |
title_fullStr |
Fault tolerance issues in nanoelectronics |
title_full_unstemmed |
Fault tolerance issues in nanoelectronics |
title_sort |
fault tolerance issues in nanoelectronics |
publisher |
University College London (University of London) |
publishDate |
2008 |
url |
http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.564522 |
work_keys_str_mv |
AT spagoccis faulttoleranceissuesinnanoelectronics |
_version_ |
1718140986641612800 |