Stanford Trip, 6-7 Apr 06

From Murray Wiki
Jump to navigationJump to search

The following people have sent e-mail saying they plan to attend:

  • Arlene Cole-Rhodes
  • Jean-Charles Delvenne
  • Michael Epstein
  • Zhipu Jin
  • Richard Murray
  • Henrik Sandberg
  • Ling Shi
  • Fei Wang

If your name isn't on this list, please send e-mail to Richard ASAP.

Travel to/from Stanford

We will need to be at Stanford by about 9 am on Thursday morning. There are two choices for getting there: leaving Wed night and staying in a hotel, or leaving on Thursday morning (very early) and driving straight to campus. Please edit one of the sections below and add your name to the appropriate list, so that we can make plan reservations:

Leaving Wednesday

  • Will leave Caltech around 6:30 pm, arriving at Stanford by 10 pm
  • Ling Shi
  • Fei Wang

Leaving Thursday

  • Will leave Caltech at 6:00 am arriving at Stanford by 9 am
  • Richard Murray
  • Arlene Cole-Rhodes
  • Zhipu Jin
  • Michael Epstein
  • Henrik Sandberg


We'll share rooms when possible to save a bit of money. Please put your name down for one of the rooms below. Please try to share rooms with someone who is going the same day that you are.

  • Room 1: Richard (no roomate)
  • Room 2: Arlene (no roomate)
  • Room 3: Michael & Zhipu
  • Room 4: Henrik & Jean-Charles
  • Room 5: Fei Wang


  • Workshop wiki page - contains copy of the agenda, links to presentations, notes from breakouts

White paper

Themes that came up over the course of the workshop

  • Machine-to-machine wireless and metrics for control
    • Q: is it OK to separate estimation and control
    • A: when computation is cheap, yes. Advantage: LTA (Kumar)
  • Energy, delay, rate, accuracy tradeoff
    • Q: can we cast this in the framework of estimation error
    • A: rate, accuracy is understood; delay less so; energy even less
  • Spatio-temporal problems (exploit locality)
    • Seems to be driven by specific applications; abstract?
  • Minimum information requirements
  • Games, teams and learning
  • Effects of routing
  • Feedback coding

Some possible challenge problems

  • Claire's air traffic control probelm
  • 1D surveillance problem
    • 1D line of moveable sensors tracking traffic on a path
    • Goal: maintain time-accurate estimate of traffic at source
    • Properties:
    • Communication rate/quality depends on distance (model?)
    • Estimation error depends on distance
    • Would like bound on probability of missed detection
    • Would like bound on estimation error of tracked individuals
  • Notes:
    • If you remove wireless restriction, answer is well-studied (radar tracks)
  • Variations
    • Ships in a port, gliders in Monterey Bay, air traffic control?