5G networks look to a virtual future

If I had to cite one word that featured in a number of presentations at the NGMN Alliance's industry event this week on 5G, it would be "slice". Not the most technical of terms it has to be said, but it cropped up time and time again as operators and vendors sought to explain how future mobile networks would be designed to cope with the demands of the "5G era" after 2020.

Dr Håkan Andersson, head of 5G radio product strategy at Ericsson, told me that "slice" or network slicing was essentially coined as a term to replace "virtual network", to avoid confusion with existing "virtual networks" and VPNs.

In essence, a "5G slice" represents a virtual network that can be designed to support the different use cases now being proposed for 5G, enabling operators to provide different levels of network capabilities depending on demand. In this scenario, software-defined networks and network functions virtualisation (NFV) would be core components of 5G as networks become increasingly virtualised and automated to reduce costs.

As explained by Hugo Tullberg, technical manager of the METIS project at Ericsson, 5G networks will go radically faster in future to support future mobile broadband needs, while latency will also need to be much reduced to meet the needs of different applications--although the industry recognises that meeting the oft-stated 1 millisecond goal for latency is a huge undertaking. But Tullberg said 5G will also go radically slower to support low-power applications in the Internet of Things.

This "slicing" of 5G networks is designed to support the future use cases for 5G. The NGMN Alliance has already identified 25 use cases in a 5G white paper it unveiled at Mobile World Congress earlier this month.

Andersson said he was impressed with what the NGMN Alliance--whose members include 24 mobile operators from across the globe--has done so far.

"The operators have done a good job," he said, noting that the operators started with use cases and worked back from there to define the technical requirements. The alliance is also working with a number of different vendors, standards and research groups in order to drive momentum for 5G.

There is still much to do: big decisions have to be made on spectrum assignments and usage, and how all the different types of connectivity will be brought together. One question that still often fails to elicit a clear answer is whether there will be a new air interface or waveform. The general message appears to be: it's not really that simple. Indeed, Tullberg warned that discussions about a new air interface could end up as little more than a "hair-splitting contest".

Wednesdays sessions provided a little more insight, with Philippe Keryer, CSIO at Alcatel-Lucent, outlined universal filtered OFDM (UF-OFDM) as a new possible air interface for 5G. Ericsson also currently favours an LTE evolution approach that combines elements of LTE with the "new radio Nx", as Andersson termed it.

For his part, Andersson finds it very exciting that 5G is set to take the industry beyond its traditional voice and data use cases, bringing in new vertical industries and involving a whole new range of stakeholders.

"It's the end of the beginning," he said, referring to when 5G finally comes to market. "Not the beginning of the end."--Anne