DARPA begins awards for smart spectrum management

DARPA has begun making some awards under its multi-million dollar Spectrum Collaboration Challenge.

The Defense Department’s research arm is looking for ways to use machine learning to better manage the radio spectrum. The competition will be a way for companies and others to develop proofs of concept.

A $749,932 award has gone to James Neel LLC to develop a scoring methodology. The University of Michigan has received a $374,197 award for a research project.

According to DARPA’s website, it expects to spend about $3.5 million in prizes as part of the challenge. First place will receive $2 million, second place gets $1 million, and third gets $750,000.

The goal is move away from the rigid licensed bands of spectrum. Today, managing spectrum is a human-driven process that is not adaptive to supply and demand. DARPA says the current system for managing spectrum cannot exploit the full-capacity of the spectrum.

The agency hopes that the challenge will discover ways for radio networks to autonomously collaborate to optimize how the spectrum should be used moment to moment.

The solicitation for the Broad Agency Agreement went out in July. In solicitation documents, DARPA said that there may be 50 billion wireless devices in the next few years, and that by 2030, demand for wireless access could be 250 times what it is today.

The current way of managing the spectrum leaves many bands unused while other bands are overwhelmed. The potential capacity of the spectrum is being squandered, DARPA said.

In addition to the competition for the challenge, DARPA also is contracting for support for managing the challenge. That is what the award to James Neel and the University of Michigan appears to be.

More awards are likely in the coming months.

Here is the solicitation number for the competition architecture contract: DARPA-BAA-16-48.