Experiences with University-Industry and...
Transcript of Experiences with University-Industry and...
Experiences with UniversityExperiences with University--Industry and Industry and Government Consortia at UC Berkeley Government Consortia at UC Berkeley ——
A Story of Three CentersA Story of Three CentersJan M. Rabaey, Donald O. Pederson Distinguished ProfessorJan M. Rabaey, Donald O. Pederson Distinguished Professor
Director Gigascale Silicon Research Center Co-Director Berkeley Wireless Research Center
University of California at Berkeley
TU Berlin, April 13, 2005
Why Centers Are Formed?Why Centers Are Formed?
• To accomplish a specific mission– Bring a team of researchers together for a limited
period of time
• To cross boundaries between disciplines– Traditional academic infrastructure is not well
equipped for multi-disciplinary research
• To create a critical mass in funding– A broader vision and agenda raises fund-raising to
another level
• To create higher visibility
Three Case StudiesThree Case Studies• The Berkeley Wireless Research Center
(BWRC)– A focused research center between Berkeley and
industry
• The Gigascale System Research Center– A focused multi-university research center
• The Center for Information Technology Research in Support of Society (CITRIS)– A multi-disciplinary Center of Centers
Berkeley Wireless Berkeley Wireless Research CenterResearch CenterA UniversityA University--Industry Industry Collaborative Center Collaborative Center
http://bwrc.eecs.berkeley.edu
• Participating Members:– Intel Corporation
– STMicroelectronics– Infineon Technologies
– Hitachi Ltd
– Sun Microsystems– Cisco Systems
– Agilent Technologies
• Associate Members:– Cadence Design Systems– Ericsson Radio Systems– Atmel Corporation– Qualcomm Incorporated– Philips Research– NEC Corporation– Samsung Electronics– Conexant Systems– Xilinx Incorporated– Fujitsu Laboratories
• Government Contracts: DARPA, NSF, ONR, MARCO, MURI
• 60 Graduate Students, 11 Faculty• 11,000 ft2, Downtown Berkeley, 1 block from campus • Operational since Feb.1999
Berkeley Wireless Research CenterBerkeley Wireless Research CenterA Partnership of UC Researchers, Industry, and Government
BWRC Operating ModelBWRC Operating Model• Members Participate
– Best of academic and industrial research– Resident researchers, part of research team
• Research Focus– Pre-competitive: >5 years out– Explore challenges and opportunities for
Future Integrated Wireless Systems– Understand tradeoffs between various implementation architectures
with respect to performance, power and cost• Open IP
– Results move quickly to the Public Domain: http://bwrc.eecs.berkeley.edu/
• Realistic prototype/test environment– Realize proof-of-concept prototypes using rapid design
flow from algorithm to implementation– STMicroelectronics and IBM foundry
• 130 nm moving to 90 nm CMOS and SiGe BiCMOS
The BWRC Research AgendaThe BWRC Research Agenda
Range
Dat
a R
ate
1m 10m 100m 1km 10km
1Kb
10Kb
100Kb
1Mb
10Mb
100Mb
Cellular (WAN)
3G Cellular
2.5 G Cellular
802.11 (LAN)
802.1a
Bluetooth (PAN)
Sensor networks
Metropolitan
Zigbee (PAN)
The Wireless ArenaThe Wireless Arena
More bit/secMore bit/secThe quest for spectral capacity• Improving spectral utilization: MIMO• Exploring new spectrum: 60 GHz• Overlaying spectrum: UWB• Re-cycling spectrum
The quest for spectral capacityThe quest for spectral capacity• Improving spectral utilization: MIMO• Exploring new spectrum: 60 GHz• Overlaying spectrum: UWB• Re-cycling spectrum
PicoRadioubiquitous wireless• Ultra-low cost• Ultra-low power • Small size
PicoRadioPicoRadioubiquitous wirelessubiquitous wireless• Ultra-low cost• Ultra-low power • Small size
Cheaper bitsCheaper bits
Meso-scale low-cost wireless transceivers for ubiquitous wireless data acquisition that• are fully integrated
–Size smaller than 1 cm3
• minimize power/energy dissipation– Limiting power dissipation to 100 µW
enables energy scavenging
• support low data-rates (< 100 kBit/sec)• and form self-configuring, robust, ad-hoc networks containing 100’s to 1000’s of nodes
Meso-scale low-cost wireless transceivers for ubiquitous wireless data acquisition that• are fully integrated
–Size smaller than 1 cm3
• minimize power/energy dissipation– Limiting power dissipation to 100 µW
enables energy scavenging
• support low data-rates (< 100 kBit/sec)• and form self-configuring, robust, ad-hoc networks containing 100’s to 1000’s of nodes
Some Highlights:Some Highlights:Ubiquitous Embedded WirelessUbiquitous Embedded Wireless
Berkeley PicoRadio ProjectBerkeley PicoRadio Project(Profs Rabaey, Wright, Ramchandran)
64Kmemory DW8051
µc
BaseBand
SerialInterface
GPIOInterface
LocationingEngine
Neighbor List
SystemSupervisor
DLL
NetworkQueues
VoltageConv
64Kmemory DW8051
µc
BaseBand
SerialInterface
GPIOInterface
LocationingEngine
Neighbor List
SystemSupervisor
DLL
NetworkQueues
VoltageConv
Integrated SubIntegrated Sub--100 100 µµW PicoNodeW PicoNode
Baseband(mixed-signal)
RF+ Antenna
ClockGeneration
DigitalProcessor(s)
PowerSupply
NetworkSensors
Baseband(mixed-signal)
RF+ Antenna
ClockGeneration
DigitalProcessor(s)
PowerSupply
NetworkSensors
Baseband(mixed-signal)
RF+ Antenna
ClockGeneration
DigitalProcessor(s)
PowerSupply
NetworkSensors
300 µW super-regenerativeTransceiver
Integrated sensor-node
Integrated transmitter
Mixed signal baseband
Energy scavenger
60 GHz Wireless LAN System 60 GHz Wireless LAN System
5 GHz of Unlicensed Spectrum
• Objective: Enable a fully-integrated low-cost Gb/s data communication using 60 GHz band.
• Approach: Employ emerging, standard CMOS technology for the radio building blocks. Exploit antenna array for improved gain and resilience.
(Profs Niknejad, Brodersen)
Possible in today's CMOS technology!Possible in today's CMOS technology!
1 mm
1.3 mm
60 Ghz CMOS Amplifier
Demonstrated key building blocks in 130 nm. Can predict the small signal gain, compression, and non-linear mixing.Accurate modeling is possible by extending RF methodologies. Work is underway for complete integrated prototype.
60 Ghz Mixer
Window of Opportunity: Window of Opportunity: Cognitive RadiosCognitive Radios
Time (min)
Fre
quen
cy (
Hz)
Existing spectrum policy forces spectrum to behave like a fragmented
disk
Bandwidth is expensive and good frequencies are taken
Unlicensed bands – biggest innovations in spectrum efficiency
Recent measurements by the FCC in the US show 70% of the allocated spectrum is not utilized
Time scale of the spectrum occupancy varies from msecs to hours
Profs Brodersen, Niknejad, Sahai, Tse
Cognitive RadioCognitive Radio’’ss
• Cognitive radio requirements– co-exists with legacy wireless
systems– uses their spectrum resources – does not interfere with them
• Cognitive radio properties– RF technology that "listens" to
huge swaths of spectrum – Knowledge of primary users’
spectrum usage as a function of location and time
– Rules of sharing the available resources (time, frequency, space)
– Embedded intelligence to determine optimal transmission
Con
figur
able
arr
ay RF
RF
RF
Sensor
Optimizer
ReconfigurableBaseband
Wideband RF modem
SummarySummary• Center in operation for close to 6 years
– Established public domain research model– Refined industry-academia collaborative research model
• World Class Research Results:– Also efforts in pulse-based radio’s, configurable radio’s and compute
platforms, advanced coders, ad-hoc networking, paintable computing– Semi-annual research retreats: January in Monterey, June in Lake Tahoe– Website: http://bwrc.eecs.berkeley.edu/
• Established infrastructure to support top quality wireless SOC research– CAD tools and automated design flows – BEE Emulation Engine in use– Laboratory for prototype testing
• Over 60 graduate degrees earned
““Design in the LateDesign in the Late--Silicon AgeSilicon Age””Gigascale Systems Research CenterGigascale Systems Research Center
Past, Present and FuturePast, Present and FutureSRC Board Meeting SRC Board Meeting -- March 15, 05March 15, 05
Jan M. Rabaey, Jan M. Rabaey, D.O.PedersonD.O.Pederson Distinguished Prof.Distinguished Prof.Director GSRCDirector GSRC
Scientific CoScientific Co--Director BWRCDirector BWRC
http://http://www.www.gigascale.orggigascale.org
PrePre--siliconsilicon SiliconSilicon PostPost--siliconsilicon
19501950’’ss 20202020’’ss
S RC/F ile na me/ 3
What the FCRP Offers --
The Focus Center Research Program was established to look beyond an individual company’s vision
National research centers in semiconductor technology:
Multiple-university teams; large-scale efforts, ~$8M/center/yrLong-range research horizon; emphasis on discoveryFocus on areas where evolutionary R&D may not find solutionsContinually re-focus on the most intractable problemsCollaboration within and among Centers for ‘total’ solutionsIndustry and Government collaboration
The MARCO Focus Center ProgramThe MARCO Focus Center Program
Addressing the Roadblocks in the ITRS Addressing the Roadblocks in the ITRS for CMOS and Beyond for CMOS and Beyond
IFC - Nano wires; Expanded Opto-electronics, Power.
GSRC- Centered on Systems:- Hetero Design- Soft Systems-Test
system softwaresystem software
systemssystemssystems
structuresstructuresstructures
materialsmaterialsmaterials
physicsphysicsphysics
circuitscircuits
devices/interconnectdevices/interconnect
platform / architectureplatform / architecture
application softwareapplication software
C2S2 - Mostly Circuits:- Analog/Mixed Signal- Heterogeneous Circuits- Post-CMOS circuits
MSD - Push CMOS to limits Select Post-CMOS Tasks
FENA - Augments Post-CMOS Research in Nano materials /devices.
Courtesy MARCO, S. Thomas
systems
GSRC in the Food Chain GSRC in the Food Chain
structures
materials
physicsphysicsphysics
structuresstructures
materialsmaterials
devicesdevices
circuitscircuits
logic / architecturelogic / architecture
system softwaresystem software
application HW/SWapplication HW/SW
integrated productsintegrated products GSRC Focus is GSRC Focus is ““SystemsSystems””““The design, verification, and test of complex, heterogeneous embedded systems-on-a-chip/package considering the realities and the challenges of the late-silicon age.”
The Bifurcation of The MarketsDiverging technology roadmaps
The Design Introduction ChallengeComplexity!!Proliferating design tool, mask, and fab cost
The Physics And Manufacturing ChallengesPower and energy as a limiting factor to integrationNano-scaling leads to uncertainty and reduces reliabilityMixed-signal design under severe stressTrue embedded systems integration requires mixed-everything
ReliableSystems
Leader:Todd Austin
The GSRC The GSRC Research ThemesResearch Themes
Power-AwareSystems
Leader:Janie Irwin
EmbeddedVerification
Leader:Karem Sakallah
Sharad Malik
EmbeddedSelf-Test
Leader:Tim Cheng
HeteroSystems
Leader:AlbertoSangiovanni-Vincentelli
System-LevelRoadmap
Leader:Andrew Kahng
DesignDrivers
Leader:Jan Rabaey
Soft Systems Leaders:Wen-Mei HwuKurt Keutzer
GSRC: Our TeamGSRC: Our Team
Alberto SVBerkeley
Wen-Mei HwuIllinois
Kurt KeutzerBerkeley
Janie IrwinPenn State
Todd AustinMichigan
Karem SakallahMichigan
Andrew KahngSan Diego
Tim ChengSanta Barbara
Jan RabaeyBerkeley
Sharad MalikPrinceton
Ken LutzBerkeley
Executive TeamExecutive Team
StanfordU.C. Berkeley
U.C. San Diego
MIT
Carnegie Mellon
U Illinois
Georgia Tech
U.C. Santa BarbaraU.C. Los Angeles
Caltech
Penn State
U Michigan
UT Austin
PrincetonPurdue
32 faculty32 faculty16 institutions16 institutions
Colorado
Some Success StoriesSome Success Stories
Error_L
Errorcomarator
RAZOR FFclk_del
Main Flip-Flop
clk
Shadow Latch
Q1D101
Error_L
Errorcomarator
RAZOR FFclk_del
Main Flip-Flop
clk
Shadow Latch
Q1D101
D-Cache
IF ID EX
ME
MWBRegister FileI-
Cache
3.3 mm
3 mm
Razor (Michigan):Allows circuits to operateat minimal voltage margins
PlatformDesign-Space
Export
PlatformMapping
Implementation Space
Application SpaceApplication Instance
Platform Instance
Platform
Metropolis (Berkeley):An integrated environmentfor formal platform-based design
128 128 128 128
•LPF•Sampling Time Estimation•Period adjustment•FFT or computing derivatives
••LPFLPF••Sampling Time EstimationSampling Time Estimation••Period adjustmentPeriod adjustment••FFT or computing derivativesFFT or computing derivatives
•Sinusoidal Jitter Extraction•Random Jitter Extraction
••Sinusoidal Jitter ExtractionSinusoidal Jitter Extraction••Random Jitter ExtractionRandom Jitter Extraction
sample periods every N cycles
sample periods sample periods every every N N cyclescycles
N
The filtered periods spectrum
The filtered periods spectrum reshaped with inverse filter transfer function
The noise population used for random jitter estimation
A
B
ABPP NSE_ANSE_tot=
Alternate Test (Georgia Tech, Santa Barbara):Enables test of embedded RF/mixed signal and serial I/O
Mescal (Berkeley, Princeton):Design and programmingof concurrent heterogeneous platforms
Moving ForwardMoving ForwardCreating Creating ““Roadmap for the LateRoadmap for the Late--Age Silicon System DesignAge Silicon System Design””
How to deal with power, variability and reliability from a systeHow to deal with power, variability and reliability from a system design perspectivem design perspective
Serves as global context for Serves as global context for allall the GSRC activitiesthe GSRC activities
Helps to consolidate efforts in the different themes (as well asHelps to consolidate efforts in the different themes (as well as crosscross--center) and create global milestonescenter) and create global milestones
2005 2010 The far beyondBeyond
Co
mp
lexi
ty
2000
Concurrency
And Flexibility
Self-Adaptivity
EmbracingRandomness
Error-resiliency
Moving ForwardMoving Forward
Launched CrossLaunched Cross--Theme Theme ““OnOn--Line XLine X”” InitiativeInitiative(X = Verification, Test, Tuning, Reliability, Resource, Power an(X = Verification, Test, Tuning, Reliability, Resource, Power and Leakage Management)d Leakage Management)
GSRC with its broad expertise base uniquely positioned to addresGSRC with its broad expertise base uniquely positioned to address this s this topic in an integrated perspectivetopic in an integrated perspective
A Whole Range of InterA Whole Range of Inter--Related ConceptsRelated Concepts
From GSRC Workshop in Michigan (Dec 04)From GSRC Workshop in Michigan (Dec 04)
Summing it all upSumming it all up
GSRC: GSRC: ““Addressing the challenges of the late and postAddressing the challenges of the late and post--silicon age at the silicon age at the System LevelSystem Level””
CenterCenter model has led to nonmodel has led to non--competitive collaboration between competitive collaboration between researchers at different locations researchers at different locations –– ““Not research as usualNot research as usual””
Continuing the push for radical crossContinuing the push for radical cross--boundary thinking and boundary thinking and unorthodox partnering unorthodox partnering –– continuous recontinuous re--evaluation going on to evaluation going on to ensure that agenda is audacious enoughensure that agenda is audacious enough
Past efforts have laid the foundations, first results becoming Past efforts have laid the foundations, first results becoming apparent, stay tuned for the future apparent, stay tuned for the future ……..
UC BERKELEY
UC DAVIS
UC MERCED
UC SANTA CRUZ
http://www.citrishttp://www.citris--uc.orguc.org
CITRIS OriginsCITRIS Origins
• Approximately 1000 researchers, including over 200 faculty from over 50 academic departments
• Many industrial partners• Significant State, federal, and private support
($300 million over 4 years)• CITRIS will focus on IT solutions to tough, Quality-of-life
problems, which we call:Driving Applications
UC BerkeleyUC Berkeley
UC MercedUC Merced
UC DavisUC Davis
UC Santa CruzUC Santa CruzThe purpose is inter-disciplinary, collaborative research on IT solutions to grand-challenge social and commercial problems affecting the quality of life of all citizens.
• Major new initiative started in 2001 jointly with UC Berkeley, UC Davis, UC Merced, and UC Santa Cruz
CITRISCITRISUniversity of CaliforniaUniversity of California
• They shape the research by providing theframework, motivation, linkages, and foundations
Initially-selected GRAND-CHALLENGE areas of focus
Initially-selected GRAND-CHALLENGE areas of focus
2001 CITRIS DRIVING APPLICATIONS
DISASTER RESPONSEDISASTER RESPONSE
EDUCATIONEDUCATION
ENERGYENERGY
ENVIRONMENTENVIRONMENT
HEALTH CAREHEALTH CARE
TRANSPORTATIONTRANSPORTATION
CITRISCITRISUniversity of CaliforniaUniversity of California
Research Research CentersCenters
ResearchResearchFacultyFaculty
Industrial Industrial PartnersPartners
StudentsStudents
CITRIS is like an Umbrella . . . CITRIS is like an Umbrella . . . CITRIS is like an Umbrella . . .
. . . with many members and affiliates
> 20
> 40 > 200Hundreds
State State AgenciesAgencies
CITRISCITRISUniversity of CaliforniaUniversity of California
CITRIS CORPORATE SPONSORSCITRIS CORPORATE SPONSORS
Founding Corporate MembersFounding Corporate Members
New in New in ’’04:04:
Associate Corporate MembersAssociate Corporate Members
CITRIS-Affiliated Research CentersCITRIS-Affiliated Research Centers
• Berkeley Center for the Information Society (BCIS)• Berkeley Institute of Design (BID)• Berkeley Seismological Laboratory (BSL) • Center for the Built Environment (CBE)• Center for Computational Science and Engineering• Center for Environmental and Water Resources
Engineering (CEWRE)• The Center for Geotechnical Modeling (CGM)• Electronic Cultural Atlas Initiative (ECAI)• Environmental Energy Technologies Division (EETD)
of Lawrence Berkeley National Laboratory.• Experimental Social Sciences Laboratory (X-LAB)• Institute for Transportation Studies (ITS)• Nanomaterials in the Environment, Agriculture, and
Technology (NEAT)• National Center of Excellence for Aviation Operational
Research (NEXTOR)• Partners for Advanced Transit and Highways (PATH)• Pacific Earthquake Engineering Research Center (PEER)
• Berkeley Sensor and Actuator Center (BSAC)• Berkeley Wireless Research Center (BWRC)• The Biosensor Group (BSG)• The Center for Biophotonics, Science and Technology
(CBST)• Center for Hybrid Embedded Software Systems (CHESS)• The Center for Image Processing and Integrated
Computing (CIPIC)• Center for Intelligent Systems (CIS)• Computer Security Laboratory• The Gigascale Silicon Research Center (GSRC)• Microelectronics Laboratory at U.C. Berkeley• National Energy Research Computing Center (NERSC)
of Lawrence Berkeley National Laboratory.• Optical Switching and Communications Laboratory
see: see: www.citriswww.citris--uc.orguc.orgCITRISCITRIS
University of CaliforniaUniversity of California
Electrical Engineering& Computer Science
Other Engineering
Other Science
Humanities, SocialScience, Law
Distribution of CITRIS Researchersby Major Discipline
Distribution of CITRIS ResearchersDistribution of CITRIS Researchersby Major Disciplineby Major Discipline
56%17%
17%
10%
CITRISCITRISUniversity of CaliforniaUniversity of California
To the Societal Scale
From the very small
SOCIETAL-SCALE INFORMATION SYSTEMS—SISSOCIETAL-SCALE INFORMATION SYSTEMS—SIS
Building & Using Sensor Nets
MEMS = Micro-Electro-Mechanical Systems
MEMS for Sensor Nets
Scalable, Reliable,Secure Services
CITRISCITRISUniversity of CaliforniaUniversity of California
Motes: the foundations of sensor Motes: the foundations of sensor networksnetworks
February 2000February 2001
February 2002August 2001
February 2003
CITRISCITRISUniversity of CaliforniaUniversity of California
Integrated MicrosystemsIntegrated Microsystems
Enabling Large TestbedsEnabling Large Testbeds
Golden Gate Bridge Net (Profs. Fenves, Culler)Masada (Prof. Glaser)Wild Fire Monitoring (Prof. Glaser, Sitar)Botanical Gardens (Prof. Culler)
All of these applications are at various stages of design and deployment
CITRISCITRISUniversity of CaliforniaUniversity of California
EmergencyEmergencySocial Sciences, Social Sciences, the Humanities; the Humanities;
EnvironmentEnvironment
EmergencyEmergency
EnvironmentEnvironment
Deployment and the Scientific ProcessDeployment and the Scientific Process
CITRISCITRISUniversity of CaliforniaUniversity of California
Temperature vs. Time
8
13
18
23
28
33
7/7/039:40
7/7/0313:41
7/7/0317:43
7/7/0321:45
8/7/031:47
8/7/035:49
8/7/039:51
8/7/0313:53
8/7/0317:55
8/7/0321:57
9/7/031:59
9/7/036:01
9/7/0310:03
Date
Humidity vs. Time
35
45
55
65
75
85
95
Rel H
um
idit
y (
%)
101 104 109 110 111
2003, unpublished
Bottom Top
36m
34m
30m
20m
10m
CITRISCITRISUniversity of CaliforniaUniversity of California
Growing in scope:Expanded Application AreasGrowing in scope:Expanded Application Areas
EDUCATIONEDUCATION
DISASTER RESPONSEDISASTER RESPONSE
ENERGYENERGY
ENVIRONMENTENVIRONMENT
HEALTH CAREHEALTH CARE
TRANSPORTATIONTRANSPORTATION
*NEW
*NEW
2001
CITRISCITRISUniversity of CaliforniaUniversity of California
EducationEducation
Emergency Preparedness Emergency Preparedness & Home Defense& Home Defense
Energy EfficiencyEnergy Efficiency
Environmental MonitoringEnvironmental Monitoring
Health CareHealth Care
Service to the Third World Service to the Third World Using ITUsing IT
Social Science, Social Science, Humanities & BusinessHumanities & Business
TransportationTransportation
TODAY
Tes
t B
eds
Tes
t B
eds
New CITRIS HeadquartersNew CITRIS Headquarters
~35,000 sq. ft. Nanofab ~80,000 total sq. ft. (AFS)
Highly collaborative, interactive office, laboratory, teaching space
Summary: What Makes A Center Succeed?Summary: What Makes A Center Succeed?
Clearly identified charter and operational rulesClearly identified charter and operational rules
Unambiguous definition of rights and privilegesUnambiguous definition of rights and privileges
IP management rules ironed out up frontIP management rules ironed out up front
Collaboration should be rewardedCollaboration should be rewarded
Only then will the total be larger than the sum of the componentOnly then will the total be larger than the sum of the componentss
Minimize barriers for cooperationMinimize barriers for cooperation
A compelling and absorbing (set of) visionsA compelling and absorbing (set of) visions
A researcher should be excited to be part of itA researcher should be excited to be part of it
Engaged and motivating leadershipEngaged and motivating leadership
If these conditions are not met If these conditions are not met ……..
Running a center will be nothing less than herding cats
http://bwrc.eecs.berkeley.eduhttp://gigascale.orghttp://www.citris-uc.org