Printable Version of this PageHome PageRecent ChangesSearchSign In
Wireless networks are being deployed today with unprecedented ubiquity. Both
the availability of an unlicensed band in which to operate and the decreasing
cost of hardware and ease of use, has driven many organizations to wireless
networking. More than just connecting mobile users to the internet at their
workplace or residence, wireless networks are being used as back-haul for research sensor networks,
deployed as city-wide municipal networks, for both internet access and
emergency infrastructure. It's this last class, emergency networks, that my research focuses on.
One consideration when using wireless networks for
emergency communication is ease and rapidity of deployment, as well as mobility. Once deployed,
the network must be robust and fault tolerant. However, to-date, most network topologies are rigid,
difficult to move, prone to failure and require substantial sophistication to deploy and operate.

I'll start by analysing the experiences of rescue workers in New Orleans
working with the emergency networks deployed there during Hurricane Katrina. These experiences
will help guide the design of algorithms and topologies fit for use
in rugged terrain with numerous environmental hazards.

My work advances the state of the art by providing a novel mechanism for
wireless infrastructure deployment: MonkeyMesh. MonkeyMesh is a system for
rapidly deploying a wireless network of many adhoc solar-powered nodes, in
cute little backpacks (CLBs), attached to trained monkeys. I'll present
and analyze the LHM (Listen, here, Monkey!) algorithm for training as well
as the CLB architecture. Indeed, we will see that with careful training,
and a sufficient supply of monkeys, MonkeyMesh can be deployed in rural
and decimated locations, where existing networks would fail. Once deployed,
MonkeyMesh is a self-configuring fault-tolerant adhoc network which will find gateways,
setup telecommunications infrastructure, route around monkey failures, and find
feeding opportunities, all without intervention from the operator. I test my
algorithms both in simulation and in a prolonged deployment in the far eastern jungles of Somefarawayplace.

Last modified 12 December 2007 at 3:13 pm by caleb