My research work on Sensor Networks: Overview
Current
research focus has been to design distributed and parallel computing algorithms
for devices having high speed processors assuming a reliable network connection
between the devices. Smart Sensors (Mica2/Mica2dots/iMotes) are the new
generation computing devices that have gained momentum. Sensornets (Sensor
networks) have evolved and are being deployed in automotives (BMW 700 Series
with 82 processor networked via the CAN -- Car Area Network), ships (BP cargo
ships for monitoring engine vibration and temperature), habitat monitoring (Great
Duck Island, CA), Structural monitoring (Golden Gate Bridge, CA), temperature
monitoring (HP Labs data centers, Palo Alto, CA). Current drive is to apply the
parallel computing algorithms to the Sensornet to improve the performance (data throughput, interactivity, and accuracy)
of application while considering the new set of constrains posed by the Sensornets
– power consumption and channel contention. (A product of such an effort is Distributed Regression for Sensornet from
Carlos et. al. from UC Berkeley, an algorithm for efficient distributed message
passing for projecting the large quantities of data generated by sensor
networks into lower dimensional spaces.). I see myself working on the stated
problem, in not only the theoretical side of algorithm design, but also the
systems side that itself brings out interesting design issues. To be more
specific, one of the interesting research problems that I would like to work on
is to come up with a completely distributed fault-tolerant, self configurable and
scalable routing primitive for Sensornets.
XMOHAN