NEST/WEbS/motes Feedback

1
NEST/WEbS/motes FeedbackOne persons system
(network) is another persons component Alan
PerlisOne Demo Is Worth 1,000 Papers

--Dave Clark

• Gordon Bell
• www.research.microsoft.com/gbell
• Microsoft Research
• 19 June 2002

2
Everything cyberizable will be in Cyberspace and
covered by a hierarchy of computers!
Body
Continent
Region/ Intranet
Cars phys. nets
Home buildings
Campus
World
Fractal Cyberspace a network of networks of
platforms
3
Bells law of computer class formation to cover
Cyberspace

• New computer platforms emerge based on chip
  density other fundamental hardware evolution
• Computer classes require new platforms, networks,
  and cyberization
• New apps and content develop around each new
  class
• Each class becomes a vertically disintegrated
  industry based on hardware and software standards
  AFTER IT FLAILS 5-10 YEARS

4
General comments

• Déjà vu building with small platforms
• PDP-8 12 bit, 6 KB built-in A/Doptional
  memory pages 48 KBmate, a very small, low-level
  machine?
• Understand and exploit Moores Law
• Id like to see a guiding taxonomy e.g. RAID
• Platform vs Peripheral Distributed vs
  master/slave Dumb vs smart (Dumb is where the s
  are)
• Fixed known, fixed unknown vs moving
• Power vs performance constrained
• Cross-section bandwidth sample rate and size
• Time based vs asynchronous always on off the
  air
• Environment (in/out/etc.) Reliability (gt1 tree)
• Secure vs non-secure rogue noise

5
Building app downUsing Detailed User Scenarios

• Like a database model, MATLAB, higher level,
  graphical languages for dealing with/thinking
  about mote arrays
• Imagine/construct/find an app user(s)
• User specifies all the queries that are needed to
  answer their questions
• Build it

6
Are you operating at too many levels?
Sipursue/evade scenario

• Focus on appsapps to drive research
• Making a long lived mote for common apps. Why
  batteries?
• First apps will be small area, low freq, lowno
  intelligence/node, few nodes, master/slave aka
  dumb aka peripherals
• What about effector nets?

7
An industry structure would be based on
Platform/Network/Interface or Platform/Network as
a periphery

• Standards, standards, standards count
• Get well-defined layered services platform
  model. Consider a mote array to be a single
  object.
• Especially hard to start company/industry
  nothing like this infrastructure.mainframes,
  minis, PCs, PDAs, cameras, X10, etc.
• killer apps are needed
• Like replacing X10 or complementing it
• Sound field measures indoor/outdoor/intrusion
• Adding effectors would introduce other issues
• Maybe youre the infrastructure for robots

8
MoteCAD power, performance, intelligence, vs
time (Moores Law)

• Can a batteryless, x power be built?
• At time, t, what is the longevity, performance?
  (t 3,6,9 years).
• For master slave structures --Would like a
  configuration program to satisfy constraints.
• What can you do as a distributed system?
• Any way to self measure x, y, z per mote in
  various sized fixed or variable apps?

9
WINS c1998. WEbS 2002What would an industry
look like?

• Standards, standards, standards
• UC/B can take on the role to keep open and
  provide basic platform
• Source of low (near 0 cost) platforms
• Environment sensor/effector specific platforms
• Need is for many apps
• Could you get someone like Echelon (Ken Oshman)
  involved as a mentor? VC?