Coordinate
Formation in Sensor Networks |
What |
We
demonstrate that it is possible to achieve accurate localization and tracking
of a target in a randomly placed wireless sensor network composed of inexpensive
components of limited accuracy. The crucial enabler for this is a reasonably
accurate local coordinate system aligned with the global coordinates. We
present an algorithm for creating such a coordinate system without the use
of global control, globally accessible beacon signals, or accurate estimates
of inter-sensor distances. The coordinate system is robust and automatically
adapts to the failure or addition of sensors. Extensive theoretical analysis
and simulation results are presented. Two key theoretical results are: there
is a critical minimum average neighborhood size of 15 for good accuracy
and there is a fundamental limit on the resolution of any coordinate system
determined strictly from local communication. Our simulation results show
that we can achieve position accuracy to within 20% of the radio range even
when there is variation of up to 10% in the signal strength of the radios.
The algorithm improves with finer quantizations of inter-sensor distance
estimates: with 6 levels of quantization position errors better than 10%
achieved. Finally we show how the algorithm gracefully generalizes to target
tracking tasks. |
Which |
IPSN-2002
paper Journal paper |
Why |
Locations
are important for interpreting sensor data. |
Who |
Radhika
Nagpal, Howie Shrobe, Jonathan Bachrach |
How |
simulations |
When |
Dec,
2002 |
Where |
MIT
AI Lab |
And |
|