Presentation 2020-12-11
A Degradation Prediction of Circuit Delay Using A Gradient Descent Method
Seiichirou Mori, Masayuki Gondou, Yousuke Miyake, Takaaki Kato, Seiji Kajihara,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) As the risk of aging-induced faults of VLSIs is increasing, highly reliable systems require to predict when the aging-induced faults will occur and to avoid a system failure. In this work, we propose an aging prediction method for logic circuits, in which on-chip delay measurement scheme catches delay degradation that is representative aging phenomenon. The proposed method creates an initial prediction model from circuit delay with degradation simulation, accelerated life test, circuit delay measured by preliminary experiment, and predicts circuit delay of each manufactured circuit with offset correction and adaptive model updates. Experimental results by accelerated life test for sample chips show that the proposed method enables construction of a prediction model reflecting change of degradation tendency due to operating environment with less impact by manufacturing variation.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Field test / Delay measurement / Degradation prediction / Machine learning / Gradient descent
Paper # DC2020-59
Date of Issue 2020-12-04 (DC)

Conference Information
Committee DC
Conference Date 2020/12/11(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair Hiroshi Takahashi(Ehime Univ.)
Vice Chair Tatsuhiro Tsuchiya(Osaka Univ.)
Secretary Tatsuhiro Tsuchiya(Nihon Univ.)
Assistant

Paper Information
Registration To Technical Committee on Dependable Computing
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A Degradation Prediction of Circuit Delay Using A Gradient Descent Method
Sub Title (in English)
Keyword(1) Field test
Keyword(2) Delay measurement
Keyword(3) Degradation prediction
Keyword(4) Machine learning
Keyword(5) Gradient descent
1st Author's Name Seiichirou Mori
1st Author's Affiliation Kyushu Institute of Technology(Kyutech)
2nd Author's Name Masayuki Gondou
2nd Author's Affiliation Kyushu Institute of Technology(Kyutech)
3rd Author's Name Yousuke Miyake
3rd Author's Affiliation Kyushu Institute of Technology(Kyutech)
4th Author's Name Takaaki Kato
4th Author's Affiliation Kyushu Institute of Technology(Kyutech)
5th Author's Name Seiji Kajihara
5th Author's Affiliation Kyushu Institute of Technology(Kyutech)
Date 2020-12-11
Paper # DC2020-59
Volume (vol) vol.120
Number (no) DC-288
Page pp.pp.1-6(DC),
#Pages 6
Date of Issue 2020-12-04 (DC)