AT&T’s 5G system in Dallas is capable of amazing speeds—over twice as fast as we saw on Verizon in Chicago—but millimeter wave coverage and building penetration remain great challenges.
5G is hitting spectacular speeds in the Sun Belt. Testing AT&T’s new millimeter wave 5G network in a park in Dallas, Texas, we saw download speeds of up to 1.3Gbps and coverage about 600 feet from a cell site—more than double the peak speed we saw testing Verizon’s 5G network in Chicago.
But this isn’t necessarily an “AT&T is better than Verizon” situation, as much as AT&T would like it to be. AT&T showed us 5G in a very different kind of city, with very different devices, then Verizon did. Rather than proclaiming who’s the best here, AT&T filled in some puzzle pieces missing after we ran speed tests in the Windy City’s canyons.
AT&T has been going great guns with network improvements recently. Its LTE network was recently judged the fastest, and since December, it has been offering real Netgear Nighthawk 5G hotspots to actual business clients in 19 different US cities. This is an invitation-only rollout, where none of the hotspot owners are being charged a monthly fee right now. AT&T is rolling things out this way so it can learn how millimeter wave works and what businesses want to do with it, AT&T SVP of marketing Abhi Ingle said. That strategy is paying off.
Ingle said the hotspots are already being used at a “new age retail business,” at an architecture firm, in construction, and in movie production. At Magnolia Market at the Silos in Waco, the 5G bandwidth has been used to upload to social media “for customer engagement.” Construction and filmmaking businesses are transferring huge files from job sites where they previously would have needed couriers. At a Samsung factory outside Austin, they’re experimenting with almost real-time control of the factory floor.
“Our launch was so important. The amount of learning and understanding you can get from a real network with a real device, all standards-based, really informed the whole industry. We wouldn’t have been showing one gig on March 29 if we weren’t launching in December [2018.] Others wouldn’t have launched that early if we hadn’t launched in December,” AT&T SVP of wireless technology Igal Elbaz said.
AT&T’s marketing has clouded these network wins. The excellent 4G network is showing up with a misleading 5G E icon on many phones, leading people to believe it’s 5G when it isn’t. (AT&T argues that the “E” should make it clear that it’s not actually 5G, which is indeed confusing.) And when AT&T started with its limited hotspot rollout, it declared that the hotspots would have a 15GB data cap, which hasn’t been imposed on anyone. But it’s important to understand these are fixable blips. The widely mocked fake 5G indicator, hopefully, will be seen for what it is; the service plans will be adjusted. What really matters is the potential of the network.
Big Town, Big Speeds
My testing experience with AT&T was very different from the experience with Verizon, so I’m hesitant to compare them directly. Verizon just gave me a retail phone and flung me out into Chicago. AT&T surrounded me with a phalanx of network techs and took me to specific locations for testing.
Although both carriers’ base stations are made by Ericsson, AT&T’s site on a roof in relatively low-lying uptown Dallas is a completely different scene from the dense, reflection-heavy skyscraper canyons I was wandering in Chicago. The light poles Verizon mounted its Chicago sites on are often lower than surrounding buildings, filling the area with reflections and echoes. AT&T’s site, meanwhile, offered a crystal-clear, line-of-sight connection over a considerable area.
The device used to test matters a lot, too. I got those 1.3Gbps speeds with a Netgear Nighthawk 5G hotspot tethered via USB to a Lenovo laptop at 165 feet from the cell site. But at the same distance, a Samsung Galaxy S10 connected via Wi-Fi to the hotspot got only 465Mbps. That’s because there’s an effective ceiling of around 600Mbps on the currently available 802.11ac Wi-Fi system. Faster 802.11ax hotspots, like Netgear’s presumed next hotspot, will help lift that barrier, but for now, to push the limits of 5G, you need a cable.
Aside from one test against a site at AT&T’s headquarters, the AT&T 39GHz millimeter wave cell site we mostly tested against is at Klyde Warren Park, a two-block park on a decked-over highway in Dallas’ Arts District. It’s surrounded by food trucks and has a glassed-in restaurant in the park. While most of the park is a lawn, leafy trees border the edges. The cell site is on a building on a neighboring block, and we had a clear line of sight to it from everywhere we tested.
AT&T’s 39GHz is a slightly higher frequency than T-Mobile’s and Verizon’s 28GHz, but it’s widely considered to have the same propagation characteristics. AT&T has an average of 375MHz of spectrum in major metros, which means it could offer, in theory, speeds of up to about 2.4Gbps. That will be dependent on software that isn’t yet available.
One of the big differences between 4G and 5G is how much can be upgraded through software, Elbaz said. In the 4G world, you often have to replace a board in a box to upgrade a network site. That’s a slow, laborious process. In a 5G world, it will be more about code, Elbaz said.
“We are in a software world; we are in an agile world. How can we apply the same methodologies to the wireless system?” Elbaz said.
We tested at distances from 150 to 600 feet from the site. Beyond 600 feet, the hotspot started to fall off 5G back to 4G. We tried to use both the USB-tethered laptop and Wi-Fi every time, but that didn’t work out because the laptop kept overheating. Here’s the result, and it’s a lot better than I saw on Verizon in Chicago:
“What we’re seeing now is a cell edge around 200 or 250 meters, which is in line with what we thought 39GHz systems would be doing. We’re seeing good experience in those ranges, and beamforming and other capabilities will show up much faster than you think,” Elbaz said.
AT&T’s download speeds are much higher than I saw on Verizon, and latency is a bit better. I saw 16-20ms of latency on AT&T’s 5G as compared with 20-25ms on LTE. Uploads were still running over LTE; that’s an issue that everyone is having right now, which will be fixed later through software.
There are a lot of variables to explain why I got slower speeds on Verizon. Could the Verizon Moto Z3’s Mod be inherently slower than a big Netgear hotspot? Was I blocking the signal when I was holding it in my hand? Are Verizon’s Chicago towers lower power, because they’re in a dense urban environment? AT&T, of course, would like me to think it’s down to its superior network engineers. Clearly, its network engineers are good. But at best that’s one part of the equation.
The view through 5G-killing glass.
5G Under Glass
Similar to my Verizon experience, an easy way to kill the 5G signal is to go behind a window. There is a restaurant in the park, about 300 feet from the cell site, all glass-walled. Immediately outside the restaurant, I got 319Mbps via Wi-Fi. Just a few feet inside, I got 92Mbps.
Percentage-wise, this is very similar to the drop I found when going inside a Starbucks in Chicago near a Verizon 5G cell site. There, much closer to the site, Verizon speeds dropped from 600Mbps to 218Mbps with an intervening window. So you lose 64-71 percent of your speed when going through glass. AT&T’s answer to this challenge is the same as Verizon’s: better beamforming software, coming soon, will help.
While they started from different base speeds, AT&T and Verizon both lost about the same percentage of speed through glass.
Trees, on the other hand, don’t have as shattering an effect. Moving in and out of some tree cover at roughly the same distance from the cell site, speeds varied between 687-763Mbps, which I don’t consider to be a huge difference.
These trees didn’t impact 5G speeds much.
Why Not Wi-Fi?
Parks often have free Wi-Fi. My Klyde Warren Park experience answered one big question about why carriers would want to use 5G rather than Wi-Fi in a park: It’s faster, with even more capacity. 5G can be the bearer of Wi-Fi, not its replacement; Elbaz pointed out that food trucks around the park can use the 5G hotspot to offer free Wi-Fi to their customers.
Sitting with Ingle, we discussed some other reasons you’d want mmWave 5G rather than Wi-Fi in big public venues. In the future, 5G will have considerably lower latency and more guaranteed quality of service than the stuttery, wobbly Wi-Fi networks we’re used to.
“Stadiums are places where massive amounts of data are consumed. Imagine, when you get up to go to the restroom or go to a concession, watching your instant replays from multiple angles in clear, 360-degree video. Every venue owner is talking about increasing the value of being on site,” Ingle said.
Speaking with AT&T folks later, we discussed another possibility: Add more virtual experiences the worse your seats are. The folks up front at a concert get to be in the presence of the stars. But the ones way in the back, who are watching on a screen anyway? They can get access to the drummer cam.
Stadia can’t do that now because of limited bandwidth and a lack of a quality-of-service mechanism. But 5G opens up these possibilities. It’ll take a few years to work them out, but to make that happen, carriers and businesses need to start now.
“The real effect of what this has on the economy, you’ll see in five to ten years,” Ingle said.
More Than mmWave
AT&T’s hotspot still isn’t available to “normal” subscribers, only to invited businesses. The company will likely make its 5G network publicly available when the Samsung Galaxy S10 5G phone rolls out, after May 16 but before June 21. We’ll see real 5G service plans then, and hopefully, we’ll hear more about coverage. Unlike Verizon, which has listed neighborhoods, and Sprint, which has given maps, AT&T refuses to tell anyone other than its invitee hotspot users where it has laid out 5G in the 19 cities it has announced so far—and without a hotspot of my own, I couldn’t go hunting for it. I asked Elbaz why he isn’t focusing on broader coverage in a smaller number of metros.
“A broader number of cities allows us to bring the promise of 5G to more different types of cities and capabilities,” Elbaz said. “Everything that we’re starting is a starting point that allows us to scale afterward, and you’re going to see us moving very fast in 2019 to scale the deployment.”
While developers will want to get their hands on 5G as soon as possible, average users may want to take a pass on that initial 5G phone. AT&T has plans to overcome the millimeter wave’s coverage issues. Later this year, the carrier will start to flip its lower band 4G spectrum over to 5G, extending 5G over a much broader area. That won’t start out super fast to begin with, but it will get better with time as more AT&T users switch from 4G to 5G, freeing up spectrum to be converted. Using that network will require a different Samsung phone, coming out later this year.
“At first we are going to carve out some of our spectrum, with how we balance between our LTE capacity and the progress of 5G deployment and adoption of devices,” Elbaz said. “But then coexistence or dynamic spectrum sharing is kicking in. That allows you to run both LTE and NR on the same spectrum, and we will start testing this next year.”