Smart Dust Scenario 2
The second scenario is considerably more complicated: it involves a
loosely coupled network of SmartDust (possible heterogenous in terms
of sensor availability) motes which coordinate to collect data over
some area. Specific examples might be:
- Motes spread over a freeway collecting information about traffic
density and speed (using vibration, slight, and/or sound sensors).
- Motes spread over a crop collecting data about water and fertilizer
distribution. Using humidity, temperature, or chemical sensors.
- Motes spread over a military zone collection data about troop
movement via light, vibration, and/or sound sensors.
There are a number of technical challenges with these applications;
namely:
- Identifying nearby motes and their relative locations
- Using relative locations to get a global picture of mote location
- Exchanging information with nearby motes
- Combining information from all motes into a reasonable picture of global sensor data.
If sensor can be distributed with enough density, a diffusion network
or bloom filter may be a good model for information combination.
Determining mote location may or may not be hard. If a centralized
interrogator is available, it can probably be used to determine sensor
location (assuming it's location is known.) Without centralized
location, it may be possible to use timing delays to determine spatial
layout -- unfortunately, this requires a transmission medium which is
slow enough to have a measureable delay (not light). Existing low
power RF solutions aren't small enough.
Communication can be CCR with a central interrogator. Without, motes
will need directable lasers or broadcast RF type solutions.
Information exchange can take several forms:
- Motes broadcast their sensor information and accumulate information
from nearby motes. Each mote keeps a picture of the local state
based on some averaging function, which can be reported back to a
central interrogator.
- Motes have a specified routing protocol -- e.g. they exchange data
with specific neighbors using routing tables (this may be more
feasible if doing point-to-point optical communication).
Back to main page.
This page last updated 3.7.2000 Sam Madden