Here's the Real Truth About Verizon's 5G Network

It's very early for Verizon's 5G network. Probably too early. I spent about six hours testing out Big Red's brand-new 5G network on day two (Friday, April 5) of its launch in Chicago. It's very small, and not very reliable. But it's a beginning.

Here's the basic good news: Across 291 tests, 64 of which were performed on 5G and the rest on 4G, the Moto Z3 phone with its new 5G Moto Mod got better average download speeds than an LG V40 being tested on 4G in the same place. I saw 600Mbps peaks on the 5G device, at several different cell sites, compared with 400Mbps peaks on the 4G device. That's a 50 percent improvement, already. Hooray! Right?

This is going to get detailed and technical. If all you want is buying advice, the advice is, don't buy, at least in April of 2019. Verizon's 5G network is a few cell sites running early firmware that lacks key features. Software upgrades and further buildouts will turn this network into something you'll want, but you don't want it now.

It's unlike Verizon to launch something before its time, but the carrier seems to have been in a race with South Korea to be able to advertise the availability of the "world's first 5G-capable phone." Verizon originally scheduled its launch for April 11, but moved it back to April 5 and then April 4, seemingly as it heard more about the Korean carriers' plans. As Verizon's Moto Z3 Mod went on sale on April 3 and the Korean Samsung Galaxy S10 5G went on sale April 5, I guess Verizon gets bragging rights.

I can't help but think this is one of the rare cases where the marketers won out over the engineers at Verizon. Lots of things are missing from this network. Would it have hurt Verizon in the long haul to hold back until mid-May, when the Samsung phone is launching in the US? I don't think it would have in terms of actual sales, and it would have been a more Verizon-worthy launch. But then it wouldn't have been able to brag "First!!!111!" like some YouTube commenter hopped up on Red Bull.

This story is going to mention "future software updates" as the solution for nearly everything. Because the final revision of the 5G standard only came out in December, the carriers' infrastructure providers just haven't been able to debug and turn on all of the features they've been promising yet. The good news is, a lot of this will come as over-the-air updates.

We will head out to Chicago again in late May, and possibly again in July, to see how things are developing.

Modding Up

Right now, the only way to get on Verizon's 5G network is by attaching a $350 Moto Mod accessory to your $480 Moto Z3 phone. Evan Blass, trustworthy leaker, says the first integrated 5G phone, the Galaxy S10 5G, is coming on May 16.

The Z3 plus its Mod are bulkier than the S10 5G, and less powerful as a smartphone. But they're also going to be much cheaper. At list price, you can get 5G with the Moto Z3 for $830. The price for the Galaxy S10 5G hasn't been announced, but it will be higher than the Galaxy S10+, which starts at $999. I'm expecting $1,249.

We've reviewed the Z3 and discussed the Mod before. When you snap it on, it takes over the phone's Modem functions, and reports its battery and connection status. Some have called this "slapping a Wi-Fi hotspot onto a smartphone," which isn't exactly correct. For now, it doesn't even work as a Wi-Fi hotspot; that's coming in a future software update (the Mod currently works as a USB Modem, with a USB-C port on the bottom). In any case, the hotspot mode uses 802.11ac, not 802.11ax, which means the Wi-Fi hotspot may end up slower than the 5G connection.

The Mod's secret sauce is the USB 3 pogo pin connector on the back of the phone, which can transfer data at up to 5Gbps, faster than any early-generation 5G network. Motorola wouldn't confirm that the Mod would work with future phones, but it strongly implied it; it's also working to make it function with the existing Moto Z2 line. So this one Mod could enable an entire lineup of early Verizon 5G phones.

Mod battery life isn't great. Its runs off of its own 2,000mAh battery. The phone won't charge it, and when the Mod's battery runs dead, you're back in 4G mode. I started using my Mod at 9:30 in the morning and it was dead by 2:30 in the afternoon. I was pumping a speed test through it every two minutes and making a location request every minute, with two half-hour breaks, so call that four and a half hours of constant transmission time and location requests. You're obviously not going to be transmitting that much, and location requests burn battery, but that still feels short. Verizon and Motorola say that battery life will increase as the network improves, which is true about cellular devices in general; they last longer when they don't have to struggle.

Verizon's service plan is simple: If you're on one of the current unlimited plans, it's $10 more for unlimited, never-deprioritized 5G data. 5G data will also be free for the first three months because the network isn't truly ready.

Ugh, the Flicker

Now let's talk about the flicker.

The Moto Mod only shows its "5G UWB" icon when it's actively transferring 5G data, not when 5G service is available, so it's very difficult to know where you have 5G coverage. Background processes move data, so you'll see that 5G icon flicker on sometimes when you're in a 5G coverage area, but it won't be a reliable indicator of coverage the way all other network status icons are.

To test speeds and try to scope out coverage, I used a custom Ookla field test app to run speed tests every two minutes, interleaved, on the Motorola phone and on a separate 4G LG V40 phone (note: Ookla is owned by Ziff Davis, PCMag.com's parent company). I also had an automated process capturing a screen shot every minute and recording the location, to catch when that 5G icon flickered on.

Even with all that running, the icon would drop to 4G when I was almost certainly on 5G (like when I hadn't moved from a previous 5G test and was getting about the same speeds). Verizon needs to change this icon behavior so that it says 5G when 5G signal is available. The good news is, it could do that in a software update.

Left: 5G on a Moto Z3. Right: 4G, probably with LAA, on an LG V40.

Speeds: Just Getting Started

Verizon is using a form of 5G called millimeter wave (mmWave) that promises extremely high speeds but has very short range from towers. Speeds also drop off quickly as you walk away from a cell site, or if you go indoors. Currently, the mmWave cell sites can use up to 400MHz of spectrum. Verizon says it has 800MHz or more of spectrum in many areas, and it'll be able to use that when it gets the appropriate software update. That will dramatically increase speeds.

Eventually, mmWave should enable huge speeds. Ericsson, Verizon's infrastructure vendor in Chicago, made some grand promises in a 2018 white paper. "An NR 200MHz TDD system at 26GHz...can provide very good DL coverage to outdoor users—for example, 50-60 percent approaching 1Gbps. With larger spectrum allocations such as 400MHz, it is possible to reach multi-Gbps speeds," the company says. Verizon is, in fact, using 400MHz in Chicago, and it has beefy fiber backhaul.

In real life, I saw a maximum of 600Mbps when in theory, using 400MHz of spectrum should enable up to 2.4Gbps on mmWave 5G NR. That's all down to, guess what? Early software.

"Quite frankly, it's going to increase rapidly," Verizon network engineering VP Mike Haberman told me.

It's also important to note that not everyone is using mmWave, or exclusively using mmWave. Both AT&T and T-Mobile will be mixing mmWave with low-band 5G, which will probably have coverage and drop-off trends a lot more like LTE, and speeds 30 to 50 percent faster than LTE. Sprint is using mid-band 5G, which will have coverage similar to Sprint's 4G network and speeds in the 400 to 600Mbps range, according to the carrier.

At pretty much every 5G NR site I found, I got between 500 to 600Mbps down when near the site. That would drop quickly with distance, as I'll get to below.

This scatter chart is a lot of fun, but take it with a grain of salt. It shows the download speed of every 5G test I took, in chronological order. It's not actually that useful, but it's cool to look at. The trend between tests 10 and 30 shows me walking up to a site, walking away, back up to it again, and then away in a different direction. That cluster of high speeds on the right is a consistent 5G coverage area in the West Loop; the little group of low speeds in the middle of it is when I walked into a stone building. I'll touch on that more below.

One thing to keep an eye on right now is not the maximum speeds, but the minimum speeds. There aren't a lot of applications currently that need 1Gbps on a mobile device. But there are a lot of applications (like video streaming) that prefer 6Mbps over 2Mbps. LTE tends to have very low floors—when it's good, it's very very good, but when it's bad, it's horrid. My early 5G testing shows that 5G already has higher floors than 4G does. If that holds, then even slow 5G speeds will be a significant improvement over tough LTE situations—provided there's coverage.

Uploads and latency were much more disappointing. 5G networks are supposed to have latency below 10ms. But latency, averaging 25.7ms, was no better than LTE latency, at 25.1ms. Upload speeds, averaging 19Mbps, were lower than the average 4G upload speeds on our V40 phone, at 42Mbps.

Verizon gave an explanation you're going to get pretty familiar with: 5G uploads aren't enabled yet. Those are LTE uploads. They're waiting for a software update. Latency will also get better with time, the carrier said.

You have to understand, there are ways in which LTE is still more advanced than 5G NR, although that will change. LTE uses more complex encoding (called QAM, or quadrature amplitude modulation) and can use more antennas at once (called MIMO, or multiple-in, multiple-out). Current LTE networks support 4x4 MIMO and 256 QAM, four layers of data and eight bits per symbol.

At the moment, 5G NR supports 2x2 MIMO and 64 QAM, or two layers and six bits per symbol. But the technology supports 4x4 MIMO on handsets and even massive MIMO at the base station level, as well as more efficient types of coding. We just have to wait for devices and capabilities to evolve.

We saw this change with LTE too. When LTE networks launched in 2011, they weren't always faster than AT&T's and T-Mobile's 3.5G HSPA networks, which caused a whole mess when AT&T and T-Mobile decided to declare HSPA as 4G. But LTE had a lot more room to grow than HSPA did. While standards continued to evolve HSPA+ to a point where in theory it could achieve 336Mbps (although US carriers never went farther than the 42Mbps version), Qualcomm has 2.4Gbps LTE support now.

This evolution is going to apply to AT&T's, T-Mobile's, and Verizon's millimeter-wave networks. They all have the same versions of the same Ericsson software. It might not apply to Sprint, which is using Nokia massive MIMO base stations on lower-band spectrum; that has a whole different set of challenges.

A base station just waiting for its 5G panels.

Coverage: Pretty Limited

Verizon says it covers the Loop, West Loop, and north of the Loop in Chicago, basically up to North Avenue. I walked around for three hours and found that coverage is very spotty and doesn't extend north of Chicago Avenue. But I also found towers that looked like they hadn't quite been turned on yet, or where I suspect Verizon is about to install 5G equipment—there's a string of towers along North Avenue, every few blocks from Sedgwick to the lake, that look almost ready to go, and my 5G indicator briefly flickered on at the corner of North and LaSalle. It didn't stick reliably enough to be captured by my coverage routine or for me to run a speed test. I'm guessing that's one of a group of sites that just aren't quite on yet.

"I have sites that are ready to go, and there are more sites that are out there, deployed and ready to go," Haberman said.

I found some coverage in the Loop, lots in the West Loop, growing coverage in the River North area, and nothing north of that in Old Town.

And here's where I saw 5G:

Now, here's where I think I saw active 5G cell sites. The farthest north icon there might be a block off. I also probably missed a few; I'm obviously missing one at Chicago and LaSalle and another at Clark and Lake, I just didn't eyeball them and didn't have time to go back and check. Verizon is definitely aiming at great density here, with a site every block or two. Note how dense they are in the West Loop—it's every block. I got consistent 5G coverage on a walk through the West Loop, but I was never more than a block from a cell site.

Range: Not Quite There Yet

Verizon told me its sites have about 800 feet of range. I saw more like 300 feet of effective range, with speeds dropping below LTE levels beyond that, even though my 5G indicator would dutifully flicker on until about 450 feet. The Mod seems unable to judge when a 4G connection would be better than a 5G one, so it hangs on to 5G for dear life even if it's just eking out a few megabits. A phone should probably prefer a good lower-gen connection over a poor higher-gen one.

The chart above shows the speeds measured at 50-foot intervals walking south along Michigan Avenue from a cell site. I think what we're seeing on the LTE line is not LTE as most people know it, but LAA, the short-range LTE that uses Wi-Fi airwaves and typically drops off after 200 feet. Note that the 5G site has pretty much the same curve as the LAA site does. No, I don't know what happened with the LTE phone at 350 feet.

This chart shows the speed drop-offs at two different 5G sites, hitting one from two different angles. The gray line is walking south along Michigan Avenue from a site, keeping line of sight the whole time with only clear air between us. I got to about 350 feet before speeds got pretty low.

But what's up with those other two lines? With the orange line, there was an elevated train trestle between the 50 and 100 foot marks. That seemed to really interrupt the signal, which was then slow but stable for the next 200 feet.

The yellow line is the same cell site as the gray one, but walking at a right angle to the other, down a narrow street between large stone buildings. That also seemed to really kill speeds.

Verizon and everyone else in the industry tell me that these sort of drop-offs will be fixed by better beam-forming and beam-steering, coming, you guessed it, in a future software update.

"Beam steering and shaping are still coming along," Verizon's Haberman said. "Those could be dramatically different over the next couple of months; those cell edges are going to change."

Now before you start screaming, "My suburb/exurb/rural area has no use for sites with only 800 feet of range!" and "They're not going to put a cell site on every block in MY town!" You are correct, and they will not. This is a very center-city setup, and the carriers are already setting up these dense build-outs to handle heavy LTE traffic.

The unanswered question is whether millimeter wave will work in any other kind of environment. Verizon says it will, and that with appropriate beam steering it's gotten up to 3,000 feet of range. But I'd like to see that in action.

Hitting the Wall

5G speeds compared, indoor vs. outdoor

I got to check out another disturbing quality of millimeter wave at a Starbucks on Jefferson Street in the West Loop. Stand under the cell site, you get 600Mbps down. Go into the Starbucks, through glass, and that's cut to 218Mbps. Go around the corner and duck into the lobby of a stone building that doesn't face onto the site, and you're down to 41.5Mbps. Lower frequency bands do not have this behavior.

I'm comfortable saying that you'll get no millimeter wave signal worth its salt if you're more than one windowed wall away from line of sight to the cell site. That's really bad. Verizon says that performance will get considerably better with future software updates that include better beam-forming and beam-steering.

Are These Sites Ugly?

In terms of being ugly, you have to understand that 5G just adds two panels to an existing LTE small cell. Here's what an LTE small cell looks like, with no 5G on it:

A Verizon cell site on North Avenue that doesn't have 5G yet, but probably will soon.

Here's a close-up on a 5G cell site. You can see there's not actually much difference. If you're arguing against small cells, you lost the battle a few years ago during the LTE era. If you're just starting to argue against small cells now because they're unattractive, after not noticing them for two years, well...think about that.

This is what a Verizon cell site with a 5G panel looks like.

Here's what a 5G site looks like in urban context, in the West Loop:

Here's someone whose Photoshop skills are about on par with mine, helping to explain what you're seeing:

Verizon's 5G Report Card

We've been promised a lot of features coming with 5G. Low latency, for example, is supposed to be a key, transformative part of the new system, but it just isn't here yet. Verizon touts how it has 800MHz of spectrum in a lot of cities, but it can't use all that spectrum yet. On the other hand, up until a few weeks ago, the base stations could only handle 100MHz of spectrum.

Here's a quick "report card" of some of the key features I did and didn't see in Verizon's network:

I think this network will improve materially during summer school. Before 5G became a crazed race to turn on a network as early in the spring as possible, it sounded like things would really start to come together mid-summer. For now, Verizon, Motorola, and Ericsson have a lot of studying to do if they want to live up to what they've all been promising in their glitzy trade show presentations.

It's About the Long Haul

To understand why Verizon is investing—why everyone is investing, really—it helps to look back at 4G speed growth over the past years. We've been tracking 4G speeds in the US and Canada since 2011 (in the US) and 2013 (in Canada). Canadian networks are better than US networks, on average, so I wanted to show both growth curves.

5G, on day one, is averaging 215Mbps down. If it follows the North American 4G speed growth curves, we should have an average of 327 to 495Mbps down by 2021, and 1.2 to 1.5Gbps down by 2025. That's an average, mind you; many speeds will be higher. This is dependent on a lot of factors, of course, the central one being our government releasing a lot more spectrum for 5G over the next few years.

Really, this is only the beginning. Verizon pledges to turn on 30 cities before the end of the year. AT&T currently claims to have 12 cities working with a hotspot, which I'll check out soon. Sprint and T-Mobile are probably launching in May and June.

I've been talking to various players in the industry, and they're telling me new software is arriving for the 5G base stations every week. Things are moving extremely fast. By the time you read this, they may have changed. We're going to be driving all around the country this spring and summer checking these things out, and will report our findings in June.

For more now, see our Race to 5G feature.

What Is 5G?