System Research Perspective -through a Sensys lens

1
System Research Perspective -through a Sensys lens

• Sensys Soap Box
• David Culler
• UC Berkeley

2
A System Research formula

• Imagine a plausible future
• Create an approximation of that vision using
  technology that exists.
• Discover what is True in that world
• Empirical experience
• Bashing your head, stubbing your toe, rubbing
  your nose in it
• Quantitative measurement and analysis
• Analytics and Foundations

3
Sensys Truths

• The idle listening problem!
• Distributed algorithms (e.g, routing) with
  bounded per-node state and non-trivial change and
  uncertainty
• Claim and hold the territory weve gained
• areas of important incremental on-going
  improvement
• Low Power MAC, Multihop Forwarding, ..
• Not enough to just improve from paper to paper
• Finish the job with standards that matter
• Still not a solid dominant MAC
• IEEE 802.15.4 superframes/beacon broken, Zigbee
  punted, Wireless HART?, ISA 100
• 6LoWPAN format done RFC4944 but MAC
  interoperability island
• Where is the bar?
• 99 x 10uA 1 x 10mA gt 110 uA is baseline, 20
  uA is research goal
• 99.5-9 reliability over multiple 70 link

4
Sensys non-Truths (yet)

• Excess Redundancy
• Given the excess of redundant WSN nodes
• Selective Coverage, Capacity spreading,
• In-network processing
• At source and dest, but IN?
• Placement by blowing in the wind
• Capacity planning and placement should be easy,
  but they exist
• Points of interest
• If you want to utilize the excess push the
  technology to make it real
• SOC offerings
• Beyond the mote
• If you want to enable in-network processing,
  develop and application that actually uses it
• Beyond surge (sense and send)
• Placement assumptions should reflect placement
  reality

5
Imagine

• The Mote future is already here
• We understand many of the truths and non-truths
  in that world
• Harvest that learning in defining the next
  future
• bold, concise, revolutionary goals to shoot for
  are invaluable
• One More Element of Truth
• Complexity constraints are even more critical
  than resource constraints
• Zillions of unattended, inaccessible networked
  devices connected to the physical world engaged
  in critical functions
• The have to be simple, robust, reliable,
  resilient, provable,
• Beware Complexity

6
One more thought

• Many of the beautiful ideas that necessity has
  mothered have tremendous value in the rest of
  the tiers of computing