DARPA announces Spectrum Collaboration Challenge at IWCE 2016

The Defense Advanced Research Project Agency (DARPA), a department of the U.S. military responsible for developing emerging technologies, yesterday announced the world’s first collaborative machine-learning competition, the Spectrum Collaboration Challenge.

Immediately after the afternoon keynote speech for IWCE 2016, Paul Tilghman, the program director at DARPA focused on wireless spectrum initiatives, took the stage to announce the agency’s newest Grand Challenge. 

“We use this model in order to catalyze a community ripe for innovation,” said Tilghman. “We’ve done this in the robotics space, we’re currently doing this in the cyber world with the Cyber Grand Challenge, but I’m here today to tell you what we’re doing for the spectrum.”

Tilghman said the Spectrum Collaboration Challenge is the first of its type in the world—a collaborative machine-intelligence competition to overcome spectrum scarcity. The challenge aims to redefine the conventional spectrum-management roles of humans and machines to maximize the flow of radio-frequency signals.

Fueled by a drive for individualized connectivity and advancements in technology, a steady shift has occurred in the responsibility of radio operations away from humans and onto machines during the past 35 years, Tilghman said.

“Today, a modern radio is responsible for everything from choosing the right protocol and frequency from a predefined, pre-allocated set, to configuring the parameters of the waveform in order to get the optimal use out of the spectrum, given its current condition,” he said. “But, as our appetites continue to grow, we’re facing a challenge.”

To make current technologies work, people decide which radios can be used where and on what frequencies, how much bandwidth should be allocated, and what protocols can interact with each other.

“As the world becomes more complex, we need a way to handle this without having to delicately pre-plan every aspect of the wireless world,” said Tilghman.