The radio spectrum map of the United States is one the last bastions of organized chaos in the modern world, an adult coloring book of pastel-shaded boxes representing the divisions and subdivisions of bandwidth allocations. As it stands, there’s no real estate left for expansion, forcing the FCC to give and take as interest groups fight for space. Wouldn’t things be so much simpler if we used machines to optimize the system for us? DARPA plans to do just that: build a new system where artificial intelligence appropriates resources in real-time as needed.
The research agency just announced the Spectrum Collaboration Challenge, the latest contest in its series of Grand Challenges—this one aimed at crowdsourcing ideas for “smart systems that collaboratively, rather than competitively, adapt in real time to today’s fast-changing, congested spectrum environment… to maximize the flow of radio frequency.”
It’s an exciting prospect to look forward to considering it’s among the first attempts to patch our crumbling infrastructure. More to the point, improved radio frequency flow should result in fewer dropped signals, faster data rates, and an overall greater stability in an era where everything from automobiles to refrigerators requires radio bandwidth.
Devoid of human interference, communication systems should theoretically improve all around. For instance, researchers from Oxford University demonstrated that machine-learning could manage the vast swathes of data needed to conduct disaster relief and provide the best supply/escape routes.
DARPA’s proposal seeks to accomplish the same sort of thing, but for the radio spectrum. How exactly? By letting machine-learning algorithms sort out frequency priorities depending on their urgency, autonomously prioritizing data packets containing emergency communications or critical safety packets over say, broadcasting or amateur communications.
The idea is that wireless communication systems are more efficient if they cooperate rather than compete for bandwidth; since not all devices are simultaneously active, machine-learning should be able to coordinate bandwidth sharing.
“The primary goal… is to imbue radios with advanced machine-learning capabilities so they can collectively develop strategies that optimize the use of the wireless spectrum in ways not possible with today’s intrinsically inefficient approach of pre-allocating exclusive access to designated frequencies. The challenge is expected to take advantage of significant recent progress in the fields of artificial intelligence and machine learning.”
To properly emulate the radio bandwidth conditions needed to test potential “spectrum-sharing strategies, tactics, and algorithms,” DARPA is constructing the largest wireless testbed of its kind. This structure, called the “Colosseum,” will enable teams to carry out large-scale experiments in a controlled environment that can be configured to simulate real-life radio conditions.
Source: Gizmodo , Gizmag , and DARPA
Advertisement
Learn more about Electronic Products Magazine
