It’s so seamless you almost never notice it, but wireless communication is the foundation upon which much of modern life is built: it powers our ability to text and make calls, hail an Uber, and stream Netflix shows. With the introduction of 5G, it also promises to lower the barrier to safer self-driving cars and kick off a revolution in the internet of things. But this next leap in wireless technology will not be possible without a key ingredient: artificial intelligence.
On Wednesday, 10 teams from industry and academia competed to fundamentally change how wireless communication systems will function. The event was the sixth and final elimination round of the Spectrum Collaboration Challenge (SC2), the latest in a long line of DARPA grand challenges that have spurred development in emerging areas like self-driving cars, advanced robotics, and autonomous cybersecurity.
The challenge was prompted by the concern that the growing use of wireless technologies risks overcrowding the airwaves our devices use to talk to one another.
Traditionally, so-called radio spectrum hasn’t been allocated in the most efficient way. In the US, government agencies divvy it up into mutually exclusive frequency bands. The bands are then parceled out to different commercial and government entities for their exclusive use. While the process helps services avoid interference with one another, whoever holds the rights to a bit of spectrum rarely uses all of it 100% of the time. As a result, a large fraction of the allocated frequencies end up unused at any given moment.
The demand for spectrum has grown to the point that the wastefulness of this arrangement is becoming untenable. Spectrum is not only shared by commercial services; it also supports government and military communication channels that are critical for conducting missions and training operations. The advent of 5G networking only ups the urgency.
To tackle this challenge, DARPA asked engineers and researchers to design a new type of communication device that doesn’t broadcast on the same frequency every time. Instead, it uses a machine-learning algorithm to find the frequencies that are immediately available, and different devices’ algorithms work together to optimize spectrum use. Rather than being distributed permanently to single, exclusive owners, spectrum is allocated dynamically and automatically in real time.
“We need to put the world of spectrum management onto a different technological base,” says Paul Tilghman, a program manager at DARPA, “and really move from a system today that is largely managed by people with pen and paper to a system that’s largely managed by machines autonomously—at machine time scales.”
Over 30 teams answered the challenge at the outset of SC2 and competed over three years on increasingly harder goals. In the first phase, teams were asked to build a radio from scratch. In the second phase, they had to make their radio collaborative, so it could share information with other radio systems. In the last phase, the teams had to incorporate machine learning to make their collaborative radios autonomous.
Wednesday night, the 10 finalists went head to head in five simulated scenarios that included supporting communications for a military mission, an emergency response, and a jam-packed concert venue. Each scenario tested different characteristics, such as the reliability of the system’s service, its ability to prioritize different types of wireless traffic, and its ability to handle highly congested environments. At the end of the event, a team from the University of Florida took home the $2 million grand prize.
Right now, the winning team’s prototype is still in the very early stages and will take a while before it makes its way into our phones. DARPA hopes that SC2 will inspire increased investment and effort in continuing to refine the technology.
To have more stories like this delivered directly to your inbox, sign up for our Webby-nominated AI newsletter The Algorithm. It's free.
Yann LeCun has a bold new vision for the future of AI
One of the godfathers of deep learning pulls together old ideas to sketch out a fresh path for AI, but raises as many questions as he answers.
The dark secret behind those cute AI-generated animal images
Google Brain has revealed its own image-making AI, called Imagen. But don't expect to see anything that isn't wholesome.
The hype around DeepMind’s new AI model misses what’s actually cool about it
Some worry that the chatter about these tools is doing the whole field a disservice.
AI’s progress isn’t the same as creating human intelligence in machines
Honorees from this year's 35 Innovators list are employing AI to find new molecules, fold proteins, and analyze massive amounts of medical data.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.