東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Fault Prediction and Self-Healing Pa...

Khalil, Kasem Mohamed Ahmed.

FindBook

Google Book

Amazon

博客來

Fault Prediction and Self-Healing Paradigm for Intelligent Hardware Systems.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Fault Prediction and Self-Healing Paradigm for Intelligent Hardware Systems./
作者:	Khalil, Kasem Mohamed Ahmed.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, : 2021,
面頁冊數:	317 p.
附註:	Source: Dissertations Abstracts International, Volume: 83-01, Section: A.
Contained By:	Dissertations Abstracts International83-01A.
標題:	Computer engineering. -
電子資源:	https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28498637
ISBN:	9798522907303

Fault Prediction and Self-Healing Paradigm for Intelligent Hardware Systems.
Khalil, Kasem Mohamed Ahmed.

Fault Prediction and Self-Healing Paradigm for Intelligent Hardware Systems. - Ann Arbor : ProQuest Dissertations & Theses, 2021 - 317 p.

Source: Dissertations Abstracts International, Volume: 83-01, Section: A.

Thesis (Ph.D.)--University of Louisiana at Lafayette, 2021.

This item must not be sold to any third party vendors.

As the complexity of hardware systems grows, the failure rate, or the rate at which such systems produce faults, accelerates. Ideally, future hardware should heal faults before the faults can occur and impact a system adversely. Fault prediction is needed to identify a fault before it occurs, and this helps to heal the fault early to avoid losing data or missing some operation. Such systems, also referred to as intelligent hardware systems, are expected to revolutionize the way circuits and systems are designed, and it is the focus of this dissertation. An intelligent hardware system is expected to have mechanisms for self-healing and fault prediction. A novel mechanism for self-healing methods for Embryonic Hardware (EmHW), Network-on-Chip (NoC), and neural network is proposed. The proposed self-healing method is implemented in VHDL on Altera Arria 10 GX FPGA device. The area overhead of the proposed self-healing method for EmHW and NoC is 34% and 31%, respectively, with high reliability and the mean-time-to-failure that prove extended network age. The hardware fault prediction requires low-cost machine learning techniques. A hardware neural network optimization and reconfiguration is proposed for artificial neural networks, Long Short-Term Memory (LSTM), and Convolutional Neural Network (CNN). An Economic LSTM (ELSTM) is proposed, which saves 34% of the area and 35% of the power consumption compared to LSTM. Next, a novel Absolute Average Deviation (ADD) pooling method with very high accuracy for CNN is also. The AAD pooling achieves an accuracy of more than 98%. and has a modest 4%. It is synthesized using Synopsis in 45 nm technology and found to occupy an area of 244.466 nm2, and consume 0.31 mW of power. Two fault prediction methods are presented, and they are based on the proposed machine learning optimization methods. The proposed fault prediction methods are used for early transistor and architectural fault prediction for NoC and EmHW, using fast Fourier transform, Principal Component Analysis (PCA), Relative PCA (RPCA), ELSTM, and CNN. The proposed approaches are implemented using Tensorflow and FPGA device, and the result shows the proposed approach could predict a fault with the accuracy of more than 98%.

ISBN: 9798522907303Subjects--Topical Terms:

621879
Computer engineering.
Subjects--Index Terms:

Hardware fault prediction

Fault Prediction and Self-Healing Paradigm for Intelligent Hardware Systems.
LDR:03427nmm a2200373 4500 001 2282383
005 20211001100837.5
008 220723s2021 ||||||||||||||||| ||eng d
020 $a 9798522907303
035 $a (MiAaPQ)AAI28498637
035 $a AAI28498637
040 $a MiAaPQ $c MiAaPQ
100 1 $a Khalil, Kasem Mohamed Ahmed. $3 3561185
245 1 0 $a Fault Prediction and Self-Healing Paradigm for Intelligent Hardware Systems.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2021
300 $a 317 p.
500 $a Source: Dissertations Abstracts International, Volume: 83-01, Section: A.
500 $a Advisor: Bayoumi, Magdy A.
502 $a Thesis (Ph.D.)--University of Louisiana at Lafayette, 2021.
506 $a This item must not be sold to any third party vendors.
520 $a As the complexity of hardware systems grows, the failure rate, or the rate at which such systems produce faults, accelerates. Ideally, future hardware should heal faults before the faults can occur and impact a system adversely. Fault prediction is needed to identify a fault before it occurs, and this helps to heal the fault early to avoid losing data or missing some operation. Such systems, also referred to as intelligent hardware systems, are expected to revolutionize the way circuits and systems are designed, and it is the focus of this dissertation. An intelligent hardware system is expected to have mechanisms for self-healing and fault prediction. A novel mechanism for self-healing methods for Embryonic Hardware (EmHW), Network-on-Chip (NoC), and neural network is proposed. The proposed self-healing method is implemented in VHDL on Altera Arria 10 GX FPGA device. The area overhead of the proposed self-healing method for EmHW and NoC is 34% and 31%, respectively, with high reliability and the mean-time-to-failure that prove extended network age. The hardware fault prediction requires low-cost machine learning techniques. A hardware neural network optimization and reconfiguration is proposed for artificial neural networks, Long Short-Term Memory (LSTM), and Convolutional Neural Network (CNN). An Economic LSTM (ELSTM) is proposed, which saves 34% of the area and 35% of the power consumption compared to LSTM. Next, a novel Absolute Average Deviation (ADD) pooling method with very high accuracy for CNN is also. The AAD pooling achieves an accuracy of more than 98%. and has a modest 4%. It is synthesized using Synopsis in 45 nm technology and found to occupy an area of 244.466 nm2, and consume 0.31 mW of power. Two fault prediction methods are presented, and they are based on the proposed machine learning optimization methods. The proposed fault prediction methods are used for early transistor and architectural fault prediction for NoC and EmHW, using fast Fourier transform, Principal Component Analysis (PCA), Relative PCA (RPCA), ELSTM, and CNN. The proposed approaches are implemented using Tensorflow and FPGA device, and the result shows the proposed approach could predict a fault with the accuracy of more than 98%.
590 $a School code: 1363.
650 4 $a Computer engineering. $3 621879
650 4 $a Electrical engineering. $3 649834
650 4 $a Artificial intelligence. $3 516317
650 4 $a Design. $3 518875
650 4 $a Datasets. $3 3541416
650 4 $a Fault tolerance. $3 3561030
650 4 $a Neural networks. $3 677449
653 $a Hardware fault prediction
653 $a Machine learning
653 $a Network-on-chip
653 $a Self healing
690 $a 0464
690 $a 0544
690 $a 0800
690 $a 0389
710 2 $a University of Louisiana at Lafayette. $b Computer Engineering. $3 3348685
773 0 $t Dissertations Abstracts International $g 83-01A.
790 $a 1363
791 $a Ph.D.
792 $a 2021
793 $a English
856 4 0 $u https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28498637