Wednesday, October 13, 2010

Fault Tolerant Wireless Sensor Network

J.N. Driscoll
PID Analyzers, LLC
K. Armstrong, W. Johnson, F. Little, P. Perov, N Perova, N. Steinsultz, Z. Mostapha

Suffolk Univ. Boston, MA
J. Kochocki,
Draper Labs, Cambridge, MA
Abstract Number: 180 - 19P
Session 180 - New Developments in Analytical Instrumentation and Software
Day and Time: Sunday, March 13, 2011,

5:30PM-7PM Room 412A Georgia World Congress Center

The primary goal of this project was to set up a triplex system of sensors and wireless motes to monitor a room sized Habitat and demonstrate stability of the information in the event of single system failures of different types. A secondary goal was to have the same system of sensors in place on a Rover and wirelessly transmit the data through a system of relay motes back to the base computers in the Habitat with redundant relay motes allowing for single mote failures anywhere in the relay chain.
The sensors used are temperature, carbon dioxide (dual beam IR), oxygen (electrochemical sensor), and pressure. Crossbow motes of three different frequencies (433 MHz, 900 MHz and 2.4 GHz) are wired directly to the sensors and broadcast the data to base motes of the corresponding frequency connected by USB to laptops.
If a single computer fails, then all data for all sensors are still present in two other computers. If any individual sensor fails, two other sensors of the same type are still present in the Habitat and providing data on that particular parameter. Each sensor data is transmitted wirelessly over three separate frequency channels, so if one transmission channel fails, two others are still available for transmission of the data for that sensor. If a single mote fails at any relay point there are nearby motes of that frequency to relay the data to the base computers.
The triplex system of sensors worked well with the wireless motes inside the habitat. Fault tolerance for one fault in a particular type of sensor was demonstrated both by observing spurious wireless transmissions and also by deliberately injecting faults into the sensors. Additional details of our testing of the Habitat and the Rover system will be discussed in our presentation.

No comments:

Post a Comment