There’s a lot of talk in the industry about how LTE is going to be here for a very long time to come. The first 5G specs—the Non Standalone (NSA) version—depend on LTE, after all. But Nokia’s North American CTO Mike Murphy says while LTE will indeed be around for a while, there’s a limit to how far LTE can take a carrier, a concept he calls “LTE exhaust.”
Given the pace of mobile data usage and the emergence of things like 4K video, AR/VR, more cord cutting and so on, it’s not a stretch to say the growth rate will continue and perhaps increase.
The main thing that gives LTE the ability to carry more traffic is Massive MIMO, which operators will be rolling out this year.
The good thing about Massive MIMO is it provides the capability to increase capacity significantly—in the range of two to three times. The bad part is it’s technically and physically limited; the antenna gets more massive to a point where it’s not practical, and at some point, an operator is going to get diminished returns, he said.
“There is a limit to how far LTE will go,” Murphy told FierceWirelessTech. “We can actually fairly accurately calculate LTE exhaust.” That involves looking at an existing LTE network, advancing it as far as it can possibly go, and calculating how much traffic that network will carry. Then the question becomes: Is that enough to cover the growth rate? Once you hit the limit, the number of options starts to reduce.
Naysayers will say you can introduce more small cells or more spectrum, but small cells are expensive and getting the locations approved can take a long time, he said. And getting more spectrum is not something to take for granted.
The point is that now that 5G is getting close and Massive MIMO is on the scene, “now is the time to make this kind of calculation,” he said. If you think LTE can take you to 2025 or so, that would be mistaken. “We’re not going to get that far,” with 2023 or 2024 kind of the max for LTE. Beyond that, it’s time to move to 5G.
Of course, operators in the U.S. are well on their way to their chosen paths to 5G, and he said the FCC has done a good job in releasing high-band spectrum. Granted, millimeter wave doesn’t penetrate well, but there are ways around that. For example, millimeter wave technology can be used to get to a building and then use Wi-Fi inside the building.
But there are pros and cons for any spectrum bands, high or low. “You need both,” he said.
Then there’s midband spectrum, which he said you could almost call a sweet spot. For one, the propagation is not so bad—not as good as 600 MHz but it’s OK. At the same time, it’s quite big, and large amounts of megahertz in the spectrum band can do a lot for peak rates and capacity.
With today’s spectrum, “we’ll probably see 1.2 gigabits per second this year,” he said. But going beyond that starts to be difficult with the spectrum that carriers hold today; going to 3, 4 or 5 gigabits, practically speaking, requires more spectrum, and the mid-band is a good candidate for supporting that. “There’s great interest in that mid band,” he noted.
One of the big promises of 5G is the low latency it will bring and then the applications it will enable, and it’s not entirely clear what the business cases are going to be. But even if operators switched all their radios to 5G today, that would not give them the latency they need. For that, they actually have to make physical changes to the network architecture.
A simple way to reduce latency is to get the content closer to the end user. Of course, technology and protocols help, he said, but at the end of the day, you have to get them closer together, and that means distributing the network a lot more. Typically, the big operators in the U.S. and Canada each have eight or 10 data centers, and now they have to look at creating maybe hundreds of smaller centers, or edge clouds, to be able to reach that level of low latency that's desired for autonomous cars, for example. That means moving things to the edge.
“You don’t want to carry traffic from one side of the country to another,” he said. “You have to keep it fairly local,” and that’s a pretty big change depending on how far you take it. It’s a big discussion point and one that’s challenging because the business case is hard to judge, but it’s a decision that should be done soon due to the time it takes to make the conversion.
What is the edge? There’s no perfect answer because each customer will define it differently, he said. But in general, it means moving things closer to the end point, and then the question becomes how close.
“It depends on the use case you’re trying to achieve,” he said. For example, ultra-low latency might be used in factories, so the edge is the factory. For autonomous cars, the edge might be the highway.
“You can’t answer it completely in the abstract,” because the edge has to be tied to the use case at hand. Which might explain why there are so many different definitions of the edge depending on who you ask.