Els Parton is a scientific editor at Imec.

7 November

A team of Imec researchers, together with scientists from Rutgers University, have participated in the finals of the Darpa Spectrum Collaboration Challenge. Their idea: using AI to teach wireless devices to avoid spectral collisions.

Since Marconi sent the first wireless signal across the oceans back in 1901, nothing much has changed: a wireless link between two appliances still relies on the fact that the two parties have agreed beforehand on which frequency they’ll use to communicate. Following, the spectrum has been divided into rigidly and exclusively licensed bands, each reserved for one type of communication. Only a few narrow unlicensed bands can be used freely. These have been occupied by Wi-Fi, Bluetooth or the IoT, for example. In these unlicensed bands, traffic has exploded. In most licensed bands, in contrast, the spectrum is either underutilized or hardly used. Consider bands that are reserved for emergency communication or for sporadic bursts from satellites.

The Imec team participating in the Darpa challenge final. Credit: Steven Latré

This static frequency plan leads to problems. Just think of 4G communication that becomes unusable after an attack or of areas without 3G or 4G signal where forest fires blaze and communication is a matter of life or death. Or think of employees trying to work in train stations, coffee shops or any other place, getting frustrated with the slow Wi-Fi because of too many users in the same area.

Towards a flexible RF allocation

If we want to continue this wireless era without these hurdles, we have to rethink our spectrum usage. Software-defined radios (SDRs) could hold the solution. Indeed, SDRs can change the frequency, timing and waveform characteristics of their wireless transmissions, enabling them to move freely among the spectrum and share unused channels. Artificial intelligence and machine learning will be an indispensable part of these SDRs for smarter frequency use.

These radios don’t just need to be smart. They also – and foremost – must be collaborative and have to continuously communicate with one another to coordinate how the spectrum is used from moment to moment. AI will allow to coordinate their activity and predict the behavior of other radios that are also using the spectrum.

Such an innovative system for wireless communication is able to share the spectrum with competing radios by scanning it for free space. The main goal is to support more traffic with a quality of service that’s much better than would be possible with a fixed spectrum allocation. This, of course, without interfering with the radios that have the licensed rights to operate in those bands.

The US frequency plan with RF allocations per application

The Darpa Spectrum Collaboration Challenge

The prestigious Darpa challenges are global competitions in which researchers work on specific themes to overcome problems that affect the entire world. The team that comes up with the best solution to a specific technological challenge is awarded a cash prize, with which it can further develop its idea. A notable challenge at the end of the 1960s was to connect computers with each other. This produced the Arpanet, the predecessor of the Internet as we know it today. More recently, the Darpa Grand Challenge sparked the idea for Google’s self-driving car.

In 2017, a new challenge was launched, the Darpa Spectrum Collaboration Challenge, calling on scientists to develop the best possible wireless network system – one that also works reliably in eg crisis situations. In the competition, the organizers play out a number of scenarios where the various radios have to collaborate to run applications to the best of their abilities. One of the scenarios, for example, is set at a shopping mall with a coffee bar, a restaurant and a number of other stores. Each of them has separate communication nodes that are accessed by varying numbers of customers and data loads throughout the day. For each scenario, a great number of games is played, with each game involving a number of access points and a selection of team radios that should collaborate to achieve the best outcome.

As the only European team, IDLab – an Imec research group at the universities of Antwerp and Ghent – participated in this contest, together with scientists from US-based Rutgers University. They decided to focus on artificial intelligence. Things can go sideways when the digital information sent out by different wireless devices ‘collides’ because the devices are using the same channel on the wireless spectrum. By teaching wireless devices like smartphones to figure out what other devices are doing and predicting when they’ll use which channels, these collisions can be avoided. As a result, it’s no longer necessary to make wireless communication plans or agreements in advance – something that’s impossible in crisis situations anyway.

The interdisciplinary group of researchers predict that such AI-based solutions could be ready for use in the short term. They’re working with Antwerp’s fire department, for example, to enable them to stream live images of fires to their command vehicles. Getting those images there requires building a new wireless network. With their idea, the team finished 6th in the Darpa challenge final on 23 October.

Edited by Nieke Roos