DARPA funds 100% wireless distributed computing experiment

The Defense Advanced Research Projects Agency (DARPA) is funding a proof-of-concept study at Virginia Tech to create a distributed computing solution that operates in a totally wireless environment and includes multiple computers and handheld devices.

DARPA's interest in distributed computing stretches back more than 40 years, to the beginnings of what we now know as the Internet, when computer resources were scarce.

"Traditional wired distributed computing has been around for many years, allowing computationally intensive tasks to be performed efficiently via many, physically connected computers," said Jeffrey Reed, principal investigator for the project and Electrical Engineering professor at Virginia Tech. "Our effort will focus on developing distributed computer systems that work in a cable-free environment, which will bring a new level of flexibility to users who need to work in rapidly changing, often challenging, mobile environments."

Presently, computational power isn't scarce, but wireless integrity is, so the first part of this project will focus on demonstrating the feasibility of wireless distributed computing on DARPA's Wireless Network after Next (WNaN.)

WNaN is a communications system that senses what parts of the wireless spectrum are free (Dynamic Spectrum Access), and automatically optimizes the network topology based upon the available resources. The goal is to create robust mesh networks using mostly affordable, commercially available hardware so soldiers can be more connected but at no great monetary expense.

In August, Raytheon BBN demonstrated a WNaN network with 102 nodes offering both voice and data traffic in a simulated tactical environment.

"Depending on the conditions of the wireless channel in use by the system, a smart decision has to be made as to whether complex computations should be carried out locally --at a single location-- or in a more distributed manner," said Madhav Marathe, co-principal investigator on the project and deputy director of the Network Dynamics and Simulation Science Laboratory at Virginia Bioinformatics Institute. "This is a challenging undertaking but we will be looking at mathematically rigorous and efficient ways to make this decision-making step happen seamlessly for a wide range of mobile devices, from handheld radios used by military personnel to cell phones and remotely controlled vehicles."