Please use this identifier to cite or link to this item: http://scholarbank.nus.edu.sg/handle/10635/77741
Title: ULTRA ENERGY-EFFICIENT SUB-/NEAR-THRESHOLD COMPUTING: PLATFORM AND METHODOLOGY
Authors: ZHAO WENFENG
Keywords: Sub-/Near-threshold, energy efficiency, ultra-low voltage, digital VLSI
Issue Date: 3-Jan-2014
Citation: ZHAO WENFENG (2014-01-03). ULTRA ENERGY-EFFICIENT SUB-/NEAR-THRESHOLD COMPUTING: PLATFORM AND METHODOLOGY. ScholarBank@NUS Repository.
Abstract: Ultra-Low Voltage (ULV) sub-/near-threshold operation has been demonstrated as a feasible solution for extremely energy-constrained applications. It brings opportunities for realizing the next-geneation computing devices and platforms. However, ULV circuits are faced with many design challenges and existing solutions are less optimal for energy minimization. In this thesis, we study the design and optimization methodology for ULV circuits. First, in order to reduce the excessive design margins and to boost the circuit performance at ultra-low voltage regime, we present a surrogate modeling based statistical static timing analysis for ULV timing closure and self-body biasing technique for performance enhancement. Second, we present several design techniques to improve the performance of crucial ULV circuit building blocks, such as wide conversion range level shifter, energy-efficient intra-cell mixed-Vth standard cell library and energy-efficient gain-cell based embedded DRAM. Finally, we propose a 280nW nearly all-digital ULV temperature sensor to eliminate the power-hungry frequency reference in conventional temperature sensor designs.
URI: http://scholarbank.nus.edu.sg/handle/10635/77741
Appears in Collections:Ph.D Theses (Open)

Show full item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 
ZhaoWenfeng.pdf5.1 MBAdobe PDF

OPEN

NoneView/Download

Page view(s)

169
checked on Dec 8, 2018

Download(s)

160
checked on Dec 8, 2018

Google ScholarTM

Check


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