Verizon 5G reach is perhaps twice as far as expected, including with obstacles (NLOS.) That's the independent conclusion of Michael Thelander and team testing two live Verizon cells in Houston,
"Who would have thought a millimeter wave signal in an area 100% blocked from the serving cell tower by the surroundings would still be capable of supporting good data speeds?" they exclaim. "Millimeter wave signals are far more resilient than we expected, even at distances exceeding several thousand feet. Tree foliage, passing school buses, buildings, parked cars, balding heads, and glass impacted the received signal, but the resultant signals were still capable of delivering meaningful data rates – thanks in part to the 400 MHz radio channel. Verizon can deploy 800 MHz channels in some markets." The results were generally good up to 1,000 meters and behind buildings.
At MWC Barcelona next week, there will be dozens of analyses on mmWave costs and buildout requirements. Nearly all of them will assume reach of 200-300 meters. That would require at least hundreds of thousands of cells to cover the U.S. and probably millions.
They probably are wrong.
Huawei's $B 5G chipset delivers a "peak" of 2.0 gig download at 3.5 GHz; Qualcomm's Snapdragon X24 LTE 4G "peaks" at 2 Gig.
"Who needs fiber?" everyone was asking. Grahame Lynch predicted three years ago that wireless would be fast enough that the fiber broadband network was unnecessary. It's not that simple, but the new wireless certainly will be enough for many.
The $B Balong 5G01 is a remarkable achievement and big news. The release is below: It's a mainstream, 3GPP standard chip that's very similar to the advanced chips of leaders Qualcomm, Intel, and Apple. The processor cores are ARM. They showed off working routers for both < 6 GHz and mmWave. There are different chips for the two frequency ranges. These are first generation chips; second generation, due later in 2018, will be smaller. They expect routers for sale this year and phones in 2019.
Verizon wants 25 ms switchover for large customers needing reliability.
Verizon is planning at least hundreds of thousands of 5G mmWave cells, many of which will be backhauled using NG-PON2 from Calix. (Below.) Recently, Brett Feldman of Goldman Sachs projected Verizon is spending $20B over the next several years; from different sources, I came to a similar conclusion last year.
"Verizon's announcement is legitimizing NG-PON2," Carl Russo of Calix asserts. "It's the right choice today for 5G and customers who need the failover reliability. Because you can switch wavelengths almost instantly and upgrade easily, it's very attractive for customers with growing demand."
He added, "The cost will come down with volume and time. Today, it's the right choice for many; going forward, we expect a very wide market."
NG-PON2 requires expensive lasers of a kind that haven't been mass produced. A half-dozen companies, Adtran as well as optics specialists, are working furiously to bring down the cost.
Hans Vestberg of Verizon is passing 30M homes without raising capex.
NTT DOCOMO's very respected CTO Seizo Inoe in 2016 called high costs of 5G "a myth" in presentations at the Brooklyn 5G Summit and an IEEE conference in Kuala Lampur. He pointed out that LTE was cheaper than 3G and that much of the 5G would use existing towers and backhaul.
Vestberg, former Ericsson CEO now running Verizon's network, sees mmWave costing $200-400/home passed as they deploy to a quarter of the U.S. (Obviously, some other areas will be more expensive.) He just told a CITI investor conference 5G is
"Massively, massively cheaper than having a fiber all the way to the home to have sort of a beam in the air going to the home."
Verizon has previously said their costs to pass a home with fiber were $700 in 2007 and went down from there, presumably to $400-600. AT&T has confirmed similar as they deploy 15M lines of fiber home. I'm inferring that "massively, massively cheaper" would be $200-$400.
The "device" may be a tablet or laptop.
The headline "AT&T to Launch Mobile 5G in 2018," led me to ask whether Intel or Qualcomm is moving their mobile phone chip delivery dates up yet another year. Most telco engineers were amazed when both companies moved delivery from 2020 to 2019; could they now move up to late 2018?
It's not impossible; TSMC will be in "risk production" of 7 nm, probably required for mmWave phones. Linley estimates the mmWave chips will require 10X the transistors of 4G, raising major questions of heat and power until 7 nm. There are thousands of engineers working furiously to get the chips out the door, but this would be remarkable.
At Huawei's London MBBF, BT CEO Gavin Patterson went first, saying, "The business case isn't there." DT CTO Bruno Jacobfeuerborn, adding "For 5G, Deutsche Telekom still has no business models." FT/Orange SVP Arnaud Vamparys followed with a second agreement.
3,000 miles away, Verizon was finalizing the announcement that they would rush to deploy ~30M homes of 5G mmWave. At NGMN and elsewhere, the top tech people share freely what they know. All are working with essentially the same understanding. How can we explain the different choices?
Verizon landlines cover only about 1/4th of the U.S. They need an offering to compete with cable and DSL.
A gigabit median at 600 meters. A year ago, most experts expected mmWave to be limited to 150-250 meters. Earlier this year, Verizon mentioned good results at 400 meters. Now, Verizon is often seeing excellent results to 600 meters and further. A gigabit is common.
The result is that far fewer new cells are required. The green dots are where Verizon already has cells. As you can see, there are few red dots for new cells. The cost is coming in so low Verizon does not expect to raise their capital spending. "Management stated that overall capital expenditures would not change substantially during the 5G rollout as internal resources are shifted from the 4G to the 5G platform."
It will be fixed only in 2018, with Verizon promising to support mobile as soon as the technology is available. Intel and Qualcomm expect 5G phones in production in 2019.
Ted Rappaport of NYU has long had an alternate "channel model" that predicted higher speeds, now confirmed.
CEO McAdam in 2016 carefully said the decision hadn't been made. In fall 2017, he told Wall Street Verizon hasn't budgeted any increased spending for 5G even in 2019. He has good reason to be coy. Verizon and the Euro CEOs constantly use the (often claimed) high spending needed for 5G as a reason for regulators to be weak and governments provide subsidies. NTT DOCOMO CTO Seizo Inoe and many others planning the networks see little if any capex increase.
So I was surprised when I discovered Verizon VP Charla Rath in spring 2016 wrote, "Just as Verizon invested billions to be the first mobile carrier to deploy our world leading 4G LTE network, we plan to do the same for 5G. Sanyogita Shamsunder, Verizon’s Director of Wireless and Technology Strategy, outlined this commitment during the opening panel."
Verizon & Korea are building. NTT DOCOMO & China Mobile close. Many others mostly holding back, doing little until 2021-2023.
2019 for mobile phones, Qualcomm promises, as TSMC ramps 7 nm process to produce the chips.
Latency will be 5-10 milliseconds to the cloud controller intelligence.
1-2 milliseconds will be measured from the tower to the phone, but that's meaningless until and unless intelligence moves to the edge. That's likely very rare for a decade because there's no apparent market for 1 ms. Most "use cases" are bogus according to experts in connected cars, virtual reality, and telemedicine. Gerhard Fettweis's Tactile Internet is inspiring but years or decades away most places.
Farooq Khan says Phazr, his small company near Dallas, can compete with the giants for the millions of 5G small cells soon to deploy. Verizon is listening and is testing the Phazr equipment. Their target was to be ready by the end of 2017.
Phazr's base station is about two feet high and a foot wide. That's large enough for 384 antennas per sector, three sectors per cell. That's massive, massive MIMO. It's made possible by the very small size of mmWave antennas.
Phazr's system uses mmWave for the download but ordinary spectrum below 6 GHz for the upload. This has the advantage of requiring much less power in the mobile phone, producing less heat.
2019 Mobile 5G Phones! blazes on ads across the trade media from Qualcomm. "You’ll see 5G in 2019 for sure," Qualcomm CEO Steve Mollenkopf promises. Verizon has begun a huge, rapid deployment of mmWave, with hundreds of sites installed.
Mollenkopf himself thought it would be 2020, but new technology is allowing him to chop a year off the schedule. His main foundry, TSMC, has begun "risk production" of 7 nm chips, the next generation. It's reasonable to expect volume production in 2019. TSMC is taking delivery in 2018 on $150M EUV machines to improve critical layers. Steve is going for 2019 because he can.
Verizon will be ready to serve mobile 5G "as soon as phones are available." China Mobile and AT&T are also looking to build early. They will be expensive and probably power hogs, but a true gigabit should be possible. ("Gig LTE" will usually be 75-300 megabits to the customer, per Qualcomm.)
September 13's announcement by Lowell McAdam will go down in history as the date of the first real 5G mmWave deployment. True speeds of a gigabit are possible, rather than the 100-300 megabits more common with "Gig LTE."
We literally have hundreds of cell sites that we have up and operating and 8 markets moving to 11 markets. The global standards will be done we believe sometime this year. The chip manufacturers will have chips available in the first quarter next year for us to incorporate that are on the global standards. So this platform is not two or three years out the way it was a year ago, it's on our doorstep and I think it's going to be huge.
Fixed only at first, competing with cable. It will be 2019 or more likely 2020 before substantial numbers of mobile phones will be able to connect. Both Qualcomm and Intel have thousands of engineers at work and are optimistic about chip delivery dates. It will not be easy.
Verizon needs 5G more than any other company on Earth. Its business model is based on being better and able to charge a premium price. AT&T and especially T-Mobile have caught up in important ways. Independent testing by Open Signal has T-Mobile faster in many cities, although Verizon data disagrees.
400 meters mmWave without line of sight. Most projections of mmWave costs imply a reach about half that; longer reach would dramatically reduce the network costs. I believe the multi-billion dollars mmWave deployments at Verizon and NTT DOCOMO are going forward because of the newly demonstrated performance. Until recently, most in the industry feared that mmWave would cost so much the economics would not support even a single network. Lowell tells Morgan's Phil Cusick,
"We were at 2000 feet from the receiver in Samsung's Technology Park, we were delivering 1.8 gigs. We said, "Okay, take that truck, drive it around the backside of the building," so there is no possible way you will have a direct line of sight, 2000 feet away, it delivered 1.4 gigabytes of throughput. And the reason was that it took all the different reflections and the computers were able to process and then get that signal back up."
2017's biggest announcement - unless China does similar. NTT Docomo CEO Kazuhiro Yoshizawa plans to go "nationwide" by 2023, per the generally reliable Nikkei Asia Review. The estimate is that the three Japanese telcos (DOCOMO, KDDI, Softbank) will spend a total of ~$45B, some of it on shared infrastructure, The story has been picked up by more than a dozen publications but that's about the only details we have. (If DOCOMO is spending half of the $5G, that's about $200/person servable.)
DOCOMO has been dropping hints for several months. I'm hoping for a detailed announcement either at the June 20th shareholders meeting or the next quarterly conference call. Nikkei has a May 22 comment from Yoshiwara "We will lead the world with our 5G technology."
Top Wall Streeter follows the evidence. Verizon has said nothing about definite plans and refuses to confirm anything. I knew I was taking a risk last month when I reported they were going ahead on the largest 5G network this side of China. An error on something this big would really hurt my reputation. I went ahead because Lowell McAdam and team are experienced and very smart. Their actions are carefully planned.
Verizon would not be spending $4-7B on fiber backhaul unless they were putting $10's of billions of gear in the field. Moffett and Del Deo confirm that $1B spent of fiber implies an investment about five times greater when labor and other costs are added.
Verizon would have bought some spectrum in the auction unless they had alternate plans to add a great deal of capacity. One report has speeds on the Verizon network falling 14% since they went unlimited. Wireless growth had been falling, but with "unlimited" people aren't switching to Wi-Fi as much. The evidence is early but the trend pretty clear.
No one, not even Verizon, can offer reliable coast estimates on the 5G small cell build. There is simply no data.
Suddenly, it's possible and happening. Costs are down; performance and reach up; "commercial deployments" moved to 2019 by Verizon and AT&T; mobile may be ready for 2020. Engineers like NTT CTO Seizo Inoe are moving up their forecasts for time to revenue. Tests are going better than expected. Most of the challenges holding back big telcos are being solved.
Nothing is certain, but many top engineers are close to convinced. I believe McAdam at Verizon has decided to go full speed ahead. Six months ago, then CFO Fran Shammo told Wall Street not to put 5G into their capex models for several years. He wanted to see the test results in 2018 before going ahead. Since then, Fran's been replaced by Matt Ellis. Verizon changed 2017 test plans from a few homes near their Boston research labs to 11 cities around the country. They've made a firm commitment to a "commercial deployment" of fixed in 2019.
We have only guesses for now, but $tens of billions of decisions are being made. Until we have extensive data from the field, everyone is relying on limited lab trials and models. There are no proven facts about mmWave economics. I can't do informed reporting without some numbers to use, so I'm collecting all the data I can find. Many answers also require estimates of traffic demand, how much people are willing to pay for what technology, whether connected cars require 1 ms, 5 ms, or 10-30 ms latency, and many some unknown unknowns. Improvements necessary and welcome.
Many answers also require estimates of traffic demand, how much people are willing to pay for what technology, whether connected cars require 1 ms, 5 ms, or 10-30 ms latency, and many other datapoints not covered here. In particular, "unlimited" offerings are driving up traffic demand far above expectations even six months ago.
The gleam in Lowell McAdam's eye discussing his future network was a hint. Verizon's purchase of $1B of Corning fiber convinced me the decision had been made to go to a one-third to one-half the United States. I have no inside information; the CFO didn't have too many drinks one night and spill secrets. Rather, the evidence and the logic drove me to believe this will soon be the largest network build in the Western world. It really doesn't look like Verizon will buy Charter (or whomever.) Making looks much cheaper than buying. It's also possible the 2017-2018 testing will be severely disappointing. They can change their decision until 2019-2020, when the spending gets heavy.
The evidence below needs to be considered alongside this separate article about what's inspiring Verizon to go ahead. The technology is (almost) ready. 5-20 gigabits is working in the labs and 10,000 engineers around the world are pushing forward. VZ is being killed by cable across the unfibered half of their territory. AT&T is making noise about going out of territory with wireless to the rooftops & gigabit to the apartments. The business logic is strong.
Most previously expected deployments before 2022 to be extremely modest. The highly respected NTT DOCOMO CTO in 2015 predicted little mmWave mobile before 2022-2023. That became the common wisdom among leading engineers despite the massive pr it would be sooner. At the Brooklyn 5G, he acknowledged advances have been made and he now expects sooner. Ted Rappaport is no longer the outlier on projected volume deployment. Nothing is certain yet, but many think Onoe is on target. 3GPP, under pressure from telcos, is cooperating.
3GPP, under pressure from telcos, is cooperating. They are expediting the ridiculously named "5G New Radio Non-standalone." That's mostly a 4G LTE system with some software tweaks and the ability to use mmWave as well as Wi-Fi spectrum. The existing LTE network will handle the hard stuff,
Small cells: fiber: Exchange: Cloud RAN: Core: Internet or PSTN. I believe Verizon has made the decision to spend ~$20B and add enough small cells to cover 1/3rd to 1/2 of the U.S. The major construction will be from 2019 to 2022.
Draft for comment: more thorough and accurate version to come. Improvements and disagreements welcome.
Verizon has decided to spend $20-25,000,000,000 on 5G millimeter wave to 30-40,000,000 homes, I believe. This is not a test, trial, or “part of 11 cities” deployment. This will be the largest new network in the western world.
1/3rd to 1/2 of the U.S. will be covered. In 2019 - if all goes well - they will begin installing between 100,000 and 300,000 small cells. They hope to finish this phase in 2022 or 2023.
Each cell will have a 4G/5G upgradable radio and will immediately add capacity to LTE, right where it’s most needed. That’s why Verizon didn’t buy any spectrum in the auction; LTE & LAA will provide what’s needed, even with an “unlimited” offering.
Millimeter wave mobile communications for 5G cellular: It will work! was a revelation in 2013. Back then, most engineers believed that rain and atmosphere made mmWave spectrum would play only a modest role in mobile communications.
Looking back four years later, most of what Ted, Samini and team wrote in 2013 is confirmed. This article was triggered when I noticed then NYU grad student Mathew Samini was an author on 15 major papers in three years (below.) These included the propagation test data and a more accurate Statistical Channel Model, an essential for designing networks. (Abstract below.)
Verizon will soon turn on millimeter wave to (a very few) homes in 11 cities; the Koreans have major plans for the next 15 months. It does work. The NYU team performed tens of thousands of tests in Manhattan as well as in Brooklyn. Their data was convincing. Four years later, $billions are being spent to solve the remaining problems and start to connect hundreds of millions.
It does work! Throughout the show, two large screens showed live data from a 1 GHz mmWave + 4x4 LTE MIMO system. The results were consistently between 19 gigabits and 21 gigabits. About 1 gigabit was LTE, an example of the gigabit LTE that will be sweeping the industry in the next two years. The mmWave gear ran consistently at 19 or 20 gigabits. While Huawei said nothing about availability, I infer 2020 is the target for working production equipment for mobile.
Not one of the dozens of network engineers I've spoken to lately disagreed with Ted Rappaport's conclusion "It will work." While we have no field results on millimeter wave to consumers, there are decades of results of mmWave for backhaul, high-speed point to point wireless, and heavy military usage.
Fotis Karonis sees everything before that as "proof of concept." He's Managing Director of Mobile and also IT, where he's advancing Software Enabled Networks. He believes BT will be first in Britain, implying the other British companies will be later. He didn't say the "first commercial" cells in 2020 will be many, so his date doesn't conflict with the 2022-2024 plan for volume mmWave from NTT's Seizo Onoe.
He acknowledges that BT still has a long way to go in LTE deployment. While they claim 99% population coverage with LTE, these maps from Open Signal suggest they are not using a robust measure of actual coverage. The map above is based on tested connections to LTE. Below, the 4G and 3G maps are side by side. Note how many places 3G is working but you can't even get 4G. That's strongly suggestive of either a crowded LTE network or optimistic coverage estimates.
Karonis is strong on the need for reliability, particularly as BT adds security services as customers.
Ted Rappaport of NYU, the world's foremost mmWave researcher, disagrees with my guess that mmWave 5G will be modest until 2021-2023. My opinion is backed up by opinions from several carriers and an estimate from Ovum that fewer than 1% of lines will be millimeter wave in 2021. Ted's opinion is shared by Verizon CEO Lowell McAdam, who will deploy in Boston and probably San Francisco as soon as Verizon can get the equipment. That should be late 2017 or early 2018. Nokia and Ericsson have hundreds of engineers working on 5G mmWave.
I will be delighted to be proven wrong and see more rapid progress. Since my comment, Rappaport and team have published a seminal paper, Millimeter Wave Wireless Communications: New Results for Rural Connectivity (Abstract below.) They were able to detect a 73 GHz signal 11 kilometers away from their transmitter, a carefully aligned antenna 110 meters above average terrain. (Pictured.) They used 1 watt of transmitter power, levels similar to today's mobile phones.
A chance to watch. Press releases sometimes are as credible as politician's speeches, so I'm looking forward to watching Tuesday's live demo of a 5G network. At the Texas Wireless Summit, Arunabha Ghosh of AT&T will present Designing Ultra-Dense Networks for 5G at 9:40. At 10 a.m., AT&T will demonstrate their state of the art 5G testing. This will be one of the first public demonstrations of a 5G mmWave system. Webcast by RCR Wireless.
AT&T & Ericsson are working on phased arrays with ultra-fast beam steering, feedback-based hybrid precoding, multi-user multiple-input/multiple-output, dynamic beam tracking and beam acquisition. Beamforming and related technologies seem may be a breakthrough that extends the reach and throughput of mmWave systems. mmWave Works!, as Ted Rappaport proclaimed a few years ago. The question now is where it will prove financially practical. All those small cells and backhaul can be very expensive.
The day will mostly be devoted to connecting and automating cars, with top speakers. Robert Heath of the University of Texas is one of the organizers; he co-wrote the book on millimeter wave and is working to solve some of the remaining problems.