Imec team in top 10 of competition to free up clogged mobile networks


A team of researchers led by Flemish nanotech centre imec was among a select few to take part in the Darpa Spectrum Collaboration Challenge, which asked them to develop methods to better use radio frequencies

Only team from Europe

A few weeks ago, it was showtime in Los Angeles for a team of Flemish researchers and their American colleagues. The team led by imec – Flanders’ nanoelectronics and digital technologies research centre – made it to the final of the Darpa Spectrum Collaboration Challenge, a prestigious American tech competition.

The Defense Advanced Research Projects Agency (Darpa) has a long history of spurring innovation. Even the internet evolved from the so-called DarpaNet, a communications network developed by the American agency.

Darpa regularly challenges researchers worldwide to come up with pioneering ideas, often with major results. The technology developed by the winners of the 2005 contest focused on autonomous vehicles, for instance, laid the foundation for the self-driving car currently being developed by Google’s sister company, Waymo.

Multidisciplinary team

About three years ago, the imec researchers and colleagues from Rutgers University in New Jersey – known collectively as Team Scatter – were chosen as one of 33 research teams allowed to take part in the second edition of the Darpa Spectrum Collaboration Challenge.

Their project: To find a solution to clogged mobile networks caused by the increasing use of mobile data and inefficient allocation of radio frequencies needed to transmit it.

“I heard about the competition and realised its goals corresponded strongly with our own,” says team leader Ingrid Moerman of Ghent University’s imec cluster. “I contacted my colleagues in Antwerp and New Jersey to establish a truly multidisciplinary team.”

Our mobile spectrum is limited, and the demand for data is growing exponentially

- Professor Ingrid Moerman

The UGent imec research group specialises in data networks, while Antwerp’s imec specialists concentrate on artificial intelligence (AI). At Rutgers school of engineering, scientists contributed their skills in working with wireless hardware.

The challenge they all faced was to develop technology that prevents mobile networks from becoming saturated. The problem is partially due to our insatiable craving for data, but also by the ineffective use of the radio frequencies through which data travels to our smartphones, laptops and tablets.

Groups of radio frequencies, or frequency bands, are used for different purposes, and the total volume of these bands is called the mobile spectrum. Some frequency bands are freely available, like those for wifi, Bluetooth and the Internet of Things (IoT).

Federal governments sell licenses for other frequencies to operators, which then have the sole right to provide them. An example is telecoms network operators. But governments also keep part of the spectrum for public services, such as the military.

“Our spectrum is limited, and the demand for data is growing exponentially,” says Moerman. “So we have to use the available capacity wisely. But there are many frequencies that are hardly used by anyone. It’s like we have taps that are always open, but much of the water is just wasted because nobody needs it when it is running – and nobody is collecting it.”

Frequencies reserved for emergency calls, for example, or coming from satellites are under-used, while frequency bands for wifi, Bluetooth and IoT traffic are increasingly in demand. If the popular bands become more and more congested, it will delay or obstruct the transmission of data, just like how a wifi connection in a coffee bar slows down if too many people use it.

Sharing frequencies

“We should break down the walls between the radio frequency bands in the spectrum, so that they can share their capacity with each other,” states Moerman. “This would disperse the total volume in a smart way. For that, you of course need a well-structured collaboration between the different actors in charge of the frequencies.”

That’s why collaboration was key in Darpa’s latest contest. Participants had access to a common frequency band, which they – or actually their software systems – had to share with each other while carrying out tasks like forwarding heavy video files or supporting time-sensitive drone communication.

These tasks were part of complex game scenarios in which the contestants’ technology had to, say, co-ordinate a rescue operation during a wildfire or ensure the wifi connection at a shopping mall during peak hours. The goal for the teams was to support each other maximally while also completing their own assignments in the best possible way.

We should break down the walls between the radio frequency bands, so they can share their capacity with each other

- Ingrid Moerman

After surviving two rounds in the Spectrum Collaboration Challenge – earning them $750,000 (€681,000) both times – Team Scatter made it to the final 10. It was the only European-based team to get there.

At the end of October, the finalists battled at the Mobile World Congress in Los Angeles, inside Darpa’s high-tech test environment – aptly named “the Colosseum”. The imec team ended the competition in sixth place.

“We are happy with the result, although certain rules in the final did not fully reward our efforts to collaborate, favouring more aggressive tactics used by the other teams,” says Moerman.

In any case, Team Scatter’s methods generated much interest and ideas for a multitude of possible applications for their technology. “It could be used for common applications, such as wifi networks, but also for specific purposes like satellite communications and complex industrial schemes involving robotics.”

Photos ©Darpa: Professor Ingrid Moerman works for the IDLab, an imec research team specialised in wireless networking and machine learning (top); Darpa’s “Colosseum” provided the perfect high-tech gaming environment (centre)