5G could potentially introduce a sea change in wireless services. But, at least so far, two of the leaders in the development of the technology refuse to give up critical details on exactly how they plan to implement 5G, particularly in the mmWave spectrum.
Millimeter-wave (mmWave) spectrum is generally considered spectrum above 20GHz -- it's a spectrum that, until 5G, was often called "junk" spectrum because it was considered largely unusable for commercial offerings. That's because transmissions in the mmWave spectrum can only travel a few hundred feet, they're prone to interference, and they can't go through walls, hands or rain.
What this all means is that, if you wanted to cover a city with mmWave signals, you would need to build a cell site every 1,000 feet or so. Based on some very, very rough math, you would need around 18,000 mmWave cell sites to cover the city of Los Angeles. And if each of those sites cost $35,000 to deploy (which is the national estimate from research firm iGR for the cost of acquiring the site, equipment, power, and backhaul necessary for a mmWave small cell transmitter) covering just the city of LA alone would cost $630 million.
And even then, those signals might not actually go into any of the buildings in LA.
This is the 5G situation that is facing AT&T and Verizon.
And that's what makes us all the more curious about how these companies are deploying 5G in mmWave. Top executives aren't saying.
Mike Haberman, VP of Verizon's network engineering, declined to provide details on the operator's mmWave network deployment, including the number of sites it's working on and their configuration. He said the operator, as a rule, does not provide that kind of information -- he said Verizon also doesn't talk about its 4G sites -- because such information can be used for inaccurate calculations relating to Verizon's network performance and coverage.
Nonetheless, Verizon remains steadfastly focused on building a 5G network using mmWave spectrum. The operator said that its planned launch of 5G in 30 cities in the first half of this year would mainly use its 28GHz mmWave spectrum holdings (although Haberman did say that, eventually, Verizon would roll out 5G across all of its spectrum bands).
One thing is clear though: Verizon isn't spending a bunch of cash on its 5G buildout. The operator's planned capex for 2019 is between $17 and $18 billion, up slightly from the $16.7 billion Verizon spent on its network in 2018.
When asked about this apparent discrepancy during the operator's quarterly conference call with analysts, Verizon's CEO Hans Vestberg explained that the operator has been planning for its 5G launch for years now, so its capex won't spike in 2019 as a result. "Remember, the majority of investments in the 5G network is coming with the fiber networking we've been doing, the passive asset we're doing, the intelligent edge network design that we've been working on for years, and then, of course, you have the equipment coming at the end of it," Vestberg said, according to a Seeking Alpha transcript of his remarks. "So, we are actually -- been for several years investing into be prepared for the 5G, and this is including the capex guidance for this year for obvious reasons."
When asked about Verizon's 5G buildout, Haberman largely echoed Vestberg's comments: "If you went out to some of these big markets, and you looked at how close the cells are right now, you'd be amazed at how close they are together. And it would occur to you to say: 'That density, is that dense enough for 5G?' And the answer is, it pretty much is," he said.
It's worth noting here that network density is a critical factor in the performance of a wireless network. According to a recent report on European wireless networks from Rewheel and Tutela, "a dense LTE grid is a stronger predictor of consistently good performance (particularly handling peak traffic loads) than number of spectrum bands deployed."
Interestingly, Haberman also pointed out that critics raised the same concerns about spectrum for 2G and 3G. He said wireless providers in the US initially launched services on the 800MHz spectrum, and observers warned that they would need to build four to eight times the number of cell sites for operations in 1850-1990MHz (PCS) spectrum.
"If you look at it right now, PCS frequencies are much more valuable in downtown areas than 800MHz," Haberman explained. "Because you don't want to propagate as far, because the cells are closer together. If you're at 800MHz, it propagates too far. So what I'm saying is that the higher frequency works to your advantage if you have the right density ... If the signal dies off quicker, the sites are tighter together, they don't interfere as much. That makes the beamforming technology more helpful, because the RF gets cleaner, and I can control it more."
Concluded Haberman on the challenge of mmWave deployments: "Every negative has an advantage," he said.
AT&T's Igal Elbaz, SVP of wireless technology, offered a similar assessment on the performance of mmWave. He also declined to provide many details about the configuration of AT&T's mmWave sites, other than to say the operator is using a mix of small cells, rooftop sites and macro towers for its 5G buildout, and that the operator continues to acquire new sites for its mmWave network.
And just like Verizon, AT&T's 2019 capex doesn't reflect a massive mmWave buildout. The operator said it plans to spend around $23 billion, excluding $1 billion in FirstNet reimbursements, during 2019 on its network. Last year AT&T spent $21.3 billion and in 2017 it spent $21.6 billion. AT&T launched mobile 5G on its 39GHz spectrum in "parts" of a dozen cities last year, and plans to expand that to additional cities this year.
Further, AT&T's Elbaz said the operator's 5G buildout -- as well as the wider industry's push to 5G -- is still in its infancy. "This is a journey. And not all of the capabilities will show up in day one. Things will evolve. I just hope that, as an industry, we will evolve on the capabilities faster than in previous generations. And that, I truly believe, is what's going to happen. I see the sense of urgency in everyone, and it's fascinating to watch."
AT&T, Verizon and the rest of the nation's wireless providers are clearly working to obtain more sites for their mmWave networks. The wireless industry in general received a recent boost from the FCC, which late last year laid out guidelines for how cities should free up locations for small cell transmitters. But apparently that still isn't enough: Verizon recently launched a new advocacy effort geared toward encouraging customers to ask their local representatives to make it easier for Verizon and other companies to build out more sites for 5G.
One final, important note to add around the deployment of 5G in mmWave spectrum is that 5G initially will basically work as an add-on to 4G. So even when you're on 5G, you'll also be on 4G -- so when you roam off 5G you'll already be connected to 4G. This will prevent the disconnections and interruptions that users can experience when they move from 4G to 3G. At least according to those in the industry, 5G phones won't unduly burn through battery life despite being connected to both 5G and 4G at all times.
5G below 6GHz
The obvious difficulties around deploying widespread 5G networks in mmWave spectrum are pushing operators like AT&T and T-Mobile to move to low-band spectrum for their initial 5G buildouts. Doing so will allow them to quickly cover much more territory with a 5G signal than they could with mmWave.
Specifically, T-Mobile has said it expects to cover its entire wireless network footprint with 5G on its 600MHz spectrum by 2020. Similarly, AT&T earlier this year said it would use its own low-band spectrum to deploy 5G across its own footprint by "early" 2020. AT&T hasn't said what spectrum band it might use for its low-band 5G deployment, but those in the industry have speculated that the deployment will leverage AT&T's 850MHz spectrum -- spectrum that can cover a lot of territory and can also penetrate into buildings and other structures.
Sprint, meantime, continues to build out its mobile 5G service in its 2.5GHz holdings, spectrum that's higher than T-Mobile's 600MHz and AT&T's 850MHz but much, much lower than Verizon's 28GHz.
Why it matters
AT&T's Elbaz explained that AT&T is initially deploying its mmWave network using its 39GHz spectrum in 100MHz channels, but said the operator would expand that to four 100MHz channels in the future. He likened the move to expanding a single-lane road to a four-lane freeway, in terms of the amount of data and the speeds a wireless network can provide.
The results of AT&T's initial speed tests using these two network configurations support his view. Elbaz said that AT&T is seeing speeds of 200-300 Mbit/s on a single 100MHz channel of 5G in 39GHz spectrum, with peaks of 400 Mbit/s. However, when the operator adds in another three channels of 100MHz capacity -- essentially creating four total transmission channels, each 100MHz wide -- top speeds rise to fully 1.5 Gbit/s.
This all doesn't come as a surprise. The type of spectrum that's being used for a wireless transmission (T-Mobile's 600MHz or AT&T's 39GHz, for example) is important, but also important is the amount of spectrum (the channel size) that's being used for that transmission. In lower spectrum bands, channel sizes are often 10MHz or 20MHz wide, but in mmWave bands, channel sizes are often 100MHz wide. And when you add several channels together, speeds and capacity rise as a result.
Elbaz boasted of AT&T's ability to quickly add additional spectrum channels to its nascent 5G network. "Every capability that comes in, we are able to iterate on this quickly, and push it into the field," he said.
T-Mobile too appears to be playing with roughly the same amount of spectrum as AT&T, given its speed test results. The company boasted recently that it is seeing around 500Mbit/s speeds on its own mmWave network buildouts.
But in terms of mmWave spectrum depth, Verizon is clearly in the lead. Whereas AT&T is throwing 400MHz worth of mmWave spectrum at its 5G network, Verizon's Haberman said Verizon can throw almost 1000MHz worth of mmWave spectrum at its 5G network.
"If LTE was a two-lane highway, 5G is a 20-lane highway. That's just applying the math," Haberman said of Verizon's mmWave spectrum holdings. Verizon's 28GHz and 39GHz holdings stem from the operator's purchases of XO Communications, Straight Path and NextLink. (Verizon, and other players, could acquire more spectrum in the FCC's mmWave spectrum auctions this year.)
Verizon is already planning to make use of its massive mmWave holdings. The operator's 5G Home service launched last year using 400MHz worth of its 28GHz spectrum licenses, offering speeds of at least 300 Mbit/s with peaks of up to 1 Gbit/s. But the operator said it expects to double those speeds sometime this year when it adds another 400MHz to its 5G signals. That will basically expand Verizon's 5G transmissions into fully 800MHz worth of mmWave spectrum in the 28GHz band.
Verizon has not yet said exactly how it will sell its 5G services, so it's unclear exactly how the operator will leverage its massive mmWave spectrum holdings. But T-Mobile has already said it won't change extra for 5G. That may be because 5G in 10MHz of 600MHz won't hold a candle to 5G in 800MHz of 28GHz.
That's clearly the tack that Verizon's Haberman is taking: "LTE Advanced, in lower frequencies, is pretty darn close to [5G] NR [in terms of speeds and performance]. Other than getting a little more efficient, which we will get with NR because it is more efficient, there's really not a lot other than that," Haberman said. The key difference in 5G, he said, will be mmWave spectrum.
Either way, the market won't have long to wait to see how this dynamic will play out, since 5G smartphones, including Samsung's Galaxy S10, are scheduled to arrive at all the major carriers in the US in the coming months.