Bell Canada and an Ottawa-based research team of people from Swedish networking giant Ericsson claim they have completed the world’s first successful field testing of “AI-native” link adaptation, a technology that auto-adjusts wireless networks to keep up with demand.
Ericsson said the AI doesn’t require any new hardware because the firm’s advanced silicon can handle the features.
The approach uses “real-time” AI that accounts for factors like interference and signal quality, Bell and Ericsson said. In the field test, the companies said they saw up to 20 percent higher downstream throughput (data transfer to users’ devices), which could translate to faster downloads. They also saw a 10-percent increase in spectral efficiency, a measure of network capacity and reliability.
Ericsson provided the AI expertise as well as research and development, while Bell provided network access and its operational knowledge. Ericsson’s executive vice-president of business area networks, Per Narvinger, said this brought the company closer to providing “transformative” 5G Advanced services and, eventually, 6G.
The AI “doesn’t require any new hardware,” according to the Ericsson spokesperson, who maintained that the firm’s “advanced silicon” could handle the new features. The functionality is said to deliver cost savings by offering more capacity without the need for equipment upgrades.
The tech should become “generally available” by the end of 2025 or early 2026, the representative said. Bell added it would like to be among the first carriers to deploy the feature, but didn’t have more details to share.
RELATED: Ericsson sets up new quantum research hub in Montreal
Historically, cellular providers have accounted for increased demand by upgrading their networks with more cell sites, antennas, and wireless spectrum. They frequently deploy cells-on-wheels (COWs), or mobile cell towers, to deal with short-term spikes, such as crowds gathering for major sports events and concerts. Telecoms have explored using AI before, but mainly to identify future optimization areas for performance and efficiency.
Ericsson isn’t the only company developing live AI optimization for cell networks. In the United States (US), Nvidia is collaborating with T-Mobile, Cisco, and other partners on “AI-native” hardware and software with next-generation 6G networks in mind. However, that partnership was only unveiled in March this year, whereas Bell and Ericsson have conducted a real-world test.
The news comes as Canadian telecoms aim to reduce (though not necessarily eliminate) their dependence on the US in light of the ongoing trade war and cooling relations between the two nations. Telus recently revealed that it was building “sovereign AI factories” that, while reliant on chips from the American tech giant Nvidia, would be located in Kamloops, BC, and Rimouski, Qué.
Feature image courtesy of Frederik Lipfert on Unsplash.