Ericsson CTO: Artificial Intelligence Will Make 5G Smarter

You know all that jazz about self-driving cars, immersive virtual reality helmets, and smarter cities being enabled by the next major upgrade of nationwide wireless networks? Well, none of that is really possible unless the anticipated 5G networks develop some form of Artificial Intelligence.

Early phases of 5G wireless technology field testing are coming soon but as the new networks evolve they will need to become much smarter about devices, data traffic types, serving applications, and more, the CTO of Ericsson tells Light Reading.

How fifth-generation (5G) services will come into being is a hot topic for mobile operators and vendors right now. The official specification has not yet been defined by the 3rd Generation Partnership Project (3GPP) but lots of work is going on about what the standard will be right now. Broadly speaking, 5G networks are expected to deliver data at rates of up to a gigabit a second, which means a full HD movie could be downloaded to a compatible smartphone in minutes. (See Is This the 5G You're Looking For? for a detailed dive into the two expected phases planned for 5G deployments.)

As you might imagine, this means a lot of work needs to happen before 5G services -- which could be tens or even hundreds of times faster than 4G networks today -- actually arrive. The de facto deadline for initial 5G deployments being bandied about in the mobile industry right now is 2020, which doesn't leave a whole lot of time to get the technology right. NTT DoCoMo Inc. (NYSE: DCM) was the amongst the first carriers to suggest 2020 as an initial deployment date and much of the industry has coalesced behind that timeline. (See Getting Massive at DoCoMo's 5G Lab.)

For more on 5G, visit the dedicated 5G section here on Light Reading.

Verizon taking lead in 'pre-5G' tests in US
That's why Ericsson AB (Nasdaq: ERIC)'s CTO Ulf Ewaldsson said he is gratified by Verizon Communications Inc. (NYSE: VZ)'s surprise announcement in September that it plans to hold initial 5G field trials in 2016. (See Verizon & Partners to Field Test 5G in 2016 and The Many Faces of 5G .)

"For me, it was very, very important because it shows leadership for 5G in the US," Ewaldsson tells Light Reading.

This is particularly relevant, as the chairman of the Federal Communications Commission (FCC) recently wrote that the US needs to take action on allocating 5G spectrum if it wants to maintain its position as a "world leader" in wireless. (See FCC Chair Wants to Take 5G Higher.)

Ericsson is one of the partners for the trials with Verizon. Ewaldsson couldn't go into much detail on what his company would be doing with Verizon but he did say he expects to supply "pre-5G" equipment for the tests in 2016.

From what Light Reading is hearing separately, those trials will likely take place later in 2016, rather than sooner.

Getting smarter
Light Reading has already talked quite a bit to Ewaldsson about possibilities and practicalities of 5G, including using higher frequencies for future radios, greater speed and capacity, and how carriers need to pick up the pace on system virtualization efforts. (See Ericsson's Network Slicing: It's Far Out, Man, CEO Chat With Ulf Ewaldsson, Ericsson and Making History With 5G.)

This time he talked some more about how network design and implementation may change. Specifically, he touched on how networks will need to get smarter as the world moves over to 5G.

The idea that 5G will need context-aware networks is not new. At the very least, the new networks will need to know what kind of device they are serving (car, phone, tablet, PC, smart sensor, wearable device etc.), where the user or device is located, and what application or service they want to use (audio, video, voice, consumer data, industrial or enterprise data etc.)

The network will at least know how much bandwidth to deliver, what quality of service to apply to the data, and if the network needs to allocate resources to a particular local area. Context-aware concepts can get more complex, particularly as work on making devices more aware progresses too, but the goal is to further automate network management and make the network much more flexible for new and different services.

To facilitate these changes, machine-learning alogorithms will be part of the growing set of important technologies that underpin 5G, Ewaldsson suggests. Work on applying artifical intelligence algorithms to mobile networks is happening now. In fact, the Institute of Electrical and Electronics Engineers Inc. (IEEE) put out a call for papers this summer.

— Dan Jones, Mobile Editor, Light Reading

melao2 10/15/2015 | 12:37:21 PM
Re: AI When I read this article, the only that came to my mind was: - SKYNET
DanJones 10/12/2015 | 5:01:23 PM
Re: AI Possible. Unlikely that older networks are going to get updated though.
mhhf1ve 10/12/2015 | 4:20:26 PM
Re: AI However... are wired networks getting the same AI-treatment to optimize connections to the type of devices at the edges? I think we might want the same functionality over all networks, not just the 5G ones!
DanJones 10/12/2015 | 9:27:43 AM
AI Worth noting that AI/Machine Learning alogos are nothing new in the desktop world BTW.
Sign In