Newest DARPA Challenge: 'Shift Paradigm' With Robot Radio

The Pentagon’s Defense Advanced Research Projects Agency (DARPA) has kicked off three-year contest for the creation of an Artificial Intelligence radio tech that would de-clutter the radio spectrum, introducing a “paradigm shift” in modern technology.

Several scientific groups submitted to DARPA’s Spectrum Collaboration Challenge (SC2) effort on September 2. The contest will end in 2019 and  submissions will be tested and a winner chosen. The first three finalists will be awarded with $3.5 million to advance their work.

Since the invention of radio, management of the radio frequency spectrum has not changed significantly, primarily divided into specific areas of bandwidth, and distributed or sold to various government and/or commercial entities.

DARPA program manager Paul Tilghman explained that the major objective of the effort is to allow a radio signal to be clearly broadcast and received, in the presence of competing signals.

“The goal of the Spectrum Collaboration Challenge is to take a bunch of radio systems that have no prior knowledge of each other and say: 'Your goal is to get the best, most optimal use of the spectrum you can possibly can.' We’re going to do this in a large, open tournament where multiple teams come in and compete to bring their best designs forward,” Defense News cited Tilghman as saying.

Artificial intelligence is deemed to be the key to accomplish the task, Tilghman added.

“This idea of AI systems that can learn to collaborate with each other is really sort of a fundamentally untapped area. And very specifically, we want to know if AI can tackle this problem."

BAE Systems, a multinational defense corporation known for AI ventures, has submitted to the SC2, according to BAE sensor processing and exploitation group director Josh Niedzwiecki. He said that BAE, by submitted proposals to CS2, hopes to accelerate its radio-spectrum technologies.

"BAE Systems has been researching technology similar to the challenge to create a radio, and other sensors, that can adapt to use the spectrum more effectively using machine learning," he said to Defense News. "BAE Systems' research in this area has touched upon the different pieces of the Spectrum Collaboration Challenge, but SC2 is a unique opportunity where it pulls all the different pieces together into a single, collaborative system."

Although the submission date has passed, DARPA left an opening for other teams to participate in the challenge. Using an “open track” option, inexperienced participants working in nontraditional formats will be able to present inventions as well.

DARPA has yet to clarify what it would do with the tech, if it is successful, but a possible range of applications includes both military and commercial purposes, Tilghman said.