Caltech Center for Complex, Networked Systems

1
Caltech Center for Complex, Networked Systems

• Mani Chandy John Doyle Richard Murray
  Pietro Perona
• Shuki Bruck? Hideo Mabuchi? Mel Simon?
• David Rutledge? Bob McEliece? Jason Hickey
• (add names)
• California Institute of Technology
• 1 February 2000
• DRAFT Presentation to Steve Koonin David
  Baltimore

2
Vision Lead the development of a new theory for
networked systemsAnalysis and design of complex
systems is lacking underlying theoretical base

• Theme 1 pervasive, ubiquitous, convergent
  networking
• Heterogeneous networks merging communications,
  computing, transportation, finance, utilities,
  manufacturing, health, consumer, entertainment,
  ...
• Robustness and reliability are the dominant
  challenges
• Create a unified field theory of
  communications, computing, control, and
  intelligence

• Theme 2 post-genomics biology and medicine
  (systems biology)
• Molecular mechanisms are characterized
• Missing piece understanding of how network
  interconnection creates robust behavior from
  uncertain components in an uncertain environment
• Transition from organisms as genes, to organisms
  as networks of integrated chemical, electrical,
  fluid, and structural elements

Richard Murray CS/internet picture
3
Approach Uncertainty management for complex,
interconnected systemsCaltech is uniquely
positioned to develop fundamental insights and
tools in this area

• Key issue is uncertainty and robustness
• Management of uncertainty is central to the study
  and creation of complex, interconnected networks
  with ubiquitous computation, communications,
  control and intelligence
• These issues are common across a wide range of
  disciplines biology, computer science, fluids,
  networks
• Rigorous, deep understanding is required
• Flood of data from molecular biology
• Engineering systems of bewildering complexity
• Data and components are becoming commodities
• A fundamentally new discipline is required to go
  from information to understanding
• Current ad hoc efforts in biology, networking,
  systems engineering and software coding will not
  suffice

• Majority of current efforts are off target
• Wide variety of new initiatives on complex
  systems MIT, Stanford, Berkeley, Michigan, SFI,
  etc
• Fragmented. Inadequate focus on connection
  between theory and applications, components and
  systems.
• Requires fundamental interactions between
  scientists, engineers, and mathematicians that is
  difficult at most institutions
• Caltech is way behind, but...
• Caltech strengths allow us to take the lead
• Uniquely positioned to undertake rigorous,
  multi-disciplinary research in key areas of
  biology, computer science, fluids, networks
• Ability to attract new faculty that would not
  find a similar home in other institutions (eg,
  Arkin, Dickinson, Adelman)
• Link BSI/bioengineering with growth in CS and
  existing strengths to create internationally
  recognized center

4
Example Biological NetworksNew understanding
based on systems-level analysis
Richard M. Murray John Doyle, Mel Simon
5
Richard M. Murray Pietro Perona, Joel Burdick
Example Computation and Neural Systems
(CNS)Strong Link between CNS themes and CCNS
vision
CNS How the brain works

• Coding of info in neurons
• Neurophysics of perception
• Consciousness

?
?

• Computational models
• Artificial sensory systemsand intelligence
• Interfacing brains and computers

6
Example Quantum networksRobustness is the key
issue in quantum information technology

• Quantum interconnection and feedback
• quantum phenomena may limit the ultimate scaling
  of critical technologies
• Use active feedback to suppress unwanted quantum
  dynamics in nanoscale devices (VLSI, MEMS, etc)
• Use active feedback to stabilize and exploit
  emerging quantum behavior in nanoscale devices
  (eg, SETs)
• Formulate interconnection topologies that elicit
  robust digital behavior from few atomic degrees
  of freedom

• Quantum communication networks (MURI)
• quantum technologies could be powerful but are
  profoundly sensitive to errors and noise
• Quantum models reduce the complexity of certain
  distributed computations
• Use of quantum resources also enables new
  cryptographic protocols
• Implementations will require fault-tolerant
  storage, processing, and transmission of quantum
  information

Hideo Mabuchi MEMS micro-electromechanical
sensors SET single-electron transistor
7
Example The Lee Center

• Research in the Lee Center
• Plott economic models of networks
• Chandy wireless campus
• Doyle network modeling
• Bruck protocols and models for wireless
  worldwide web
• Effros image data compression for wireless
  networks
• McEliece turbo codes for wireless networks
• Hajimiri silicon ICs for RF Networks
• Vahala channel dropping filters for photonic
  networks
• Rutledge phased arrays for millimeter-wave
  networks
• Hickey Reliable computing in networks

• The Lee Center for Advanced Networking was
  established through a 10,000,000 donation from
  David Lee, a Caltech alumnus, who is founder and
  Chief Operating Officer for Global Crossing. The
  goal is research in the development of a
  world-wide distributed computing system that
  connects people and appliances through wireless
  and high-bandwidth wired channels and a backbone
  of computers that serve as data bases and object
  servers. This is a multi-disciplinary challenge,
  with research in modeling and protocols, circuits
  and coding, and propagation and antennas.
• Director David Rutledge

8
Connections with Traditional Caltech
StrengthsMicro-scale phenomena as a problem in
robustness of complex, uncertain systems
Richard M. Murray All

• Fluid Mechanics of turbulence and shear flows
  (Leonard, Marsden, et al)
• Model transition to turbulence as high
  selectivity, high gain operator
• Fluid Mechanics of biological systems (Gharib)
• Describe

• Global change (Wiggins, Env)
• Environment, etc

Richard Murray Picture
Richard Murray Picture
9
Richard M. Murray Working with Mark Myers to
fill out this chart. Not clear whether it
belongs here or not (or whether it is a good
thing to do, for that matter)
Industry InteractionsExplore startup company to
allow easier transition of theory to applications

• CCNS.com
• Small company dedicated to transitioning Caltech
  Center for Networked Systems (CCNS) results to
  application
• Recruit Mark Myers as director?
• Market infrastructure tools (eg Modelica
  implementations) as baseline business
• Caltech faculty as board of directors

• Potential projects
• Fuel cell modeling for International Fuel Cells
  (UTC)
• Real-time, distributed optimization and control
  for ultra-efficient, fleet operations (NIST ATP)

10
Proposal Establish a Center for Complex,
Networked Systems at CaltechBuild on Caltechs
strengths to establish leadership in complex,
interconnected systems

• 1. Establish steering committee to plan new
  center
• Work out organizational details of the center
• Recruit a director
• Begin fund raising
• Refine and articulate the vision
• 2. Expand recruiting efforts in CS and
  bioengineering
• Launch high visibility recruiting effort to
  attract leaders
• 3. Recruitment of a high-profile, permanent
  director
• Need a strong administrative and technical
  leader, eg,
• Randy Katz at UCB
• John Gutag at MIT
• Pramod Khargonekar at U. Michigan
• Make this the most exciting academic
  administrative position in the country

• Expected features of the center plan
• Highly visible commitment for new appointments in
  EAS (CS bio-engneering), with search controlled
  by steering committee
• Center director with authority and responsibility
  comparable to division chair
• Directed search for major donors
• Novel strategies for CS and bio-engineering
  education
• Benefits to Caltech
• Provide a focal point for aggressive faculty
  recruiting in CS, Bi, and EAS
• Attract students, postdocs, visitors by offering
  a uniquely multi-disciplinary environment for
  study and research
• Substantial opportunities for external funding
  (government private)

11
Event Timeline

• 4/1 Networking working meeting
• Doyle Complexity and uncertainty
• Arkin Systems biology
• Mabuchi Quantum networks
• Bruck biological networks protocols
• 4/3 M. Myers dinner (Hartford)
• 4/8 DARPA planning meeting
• Decide on agenda, speakers, etc
• 4/15 DARPA ITO visit (Sastry)
• Present Center vision and introduce research team
• 30 minute talks with open discussions
• Introduction by Koonin (?)
• 5/1 Koonin/Baltimore presentation
• This presentation
• 5/27 ERC Industry Day
• Keynote Frank Fernandez