Tuesday, April 30, 2019

U.S. Mobile Needs 5G Just to Maintain Consistent User Experience

Not every mobile market “needs” lots of additional internet access capacity now, and will need it in the next two years. The U.S. market is among the places where more capacity is an immediate issue, and the issue is whether 5G in new spectrum, spectrum refarming, smaller cells, spectrum sharing, spectrum aggregation, broader advanced 4G deployments or all the above are required.

“4G networks in the U.S. are becoming heavily loaded,” says Opensignal. So “consumers and operators alike will need 5G to relieve pressure on existing networks, otherwise the overall mobile network experience will worsen. In other words, faster speeds made possible by 5G are important, but perhaps the equally-important value is capacity reinforcement.

A shift to unlimited usage plans by all four national U.S. mobile service providers also has driven capacity demand.


“Today, 4G networks are very inconsistent in terms of speed with average connections at the busiest time of day often being half that of speeds during the quietest times of day,” Opensignal says, citing its 5G Opportunity report.

In the U.S. market, average 4G download speeds varied between 15.3 Mbps to 28.8 Mbps, depending on the time of day and congestion levels on a network.
The variation in speeds throughout the course of the day are even more pronounced in the largest cities. In Miami, average 4G download speeds were as low as 17 Mbps and as high as 43.2 Mbps over a 24-hour period. Baltimore, Chicago and New York City variations of 20 Mbps or more occurred between fastest and slowest hours.
“5G will not only provide extremely fast speeds but also a solid bedrock of capacity, to even out the consistency issues we are seeing with current 4G networks,” says Opensignal.

In other words, in some markets, 5G is a major tool for supplying the increasing demand for mobile data, which in the U.S. mobile market is growing 46 percent annually, according to Cisco.


The point is that 5G is needed, in the U.S. market, especially by AT&T and Verizon, the two networks with the most customers and therefore capacity demand to support.

U.S. Mobile Prices are Not "High"

One often hears that U.S. mobile internet access prices are among the highest in the world. As Samuel Clemens (Mark Twain) is reported to have said, there are lies, damned lies, and statistics. And cross-country internet access prices arguably are included in the sphere of such realities.

The problem with comparing prices for anything globally is that price levels are different from one country to the next. So it should not be surprising that mobile internet access prices vary as well.

Economists use a concept known as purchasing power parity to normalize for such differences, removing distortions based on the differing value of currencies (and therefore purchasing power) as well as differences in cost of living.

Once prices are measured using the PPP method, it turns out that mobile internet access prices are significantly lower than the world average in places such as Europe, the United States and the developed world in general .

As you might expect, prices are below world averages in all developed nations and U.S. household spending on communications is among the lowest in the world. That arguably is true for fixed network internet access as much as mobile access.

One sees the same trend if internet access cost is adjusted for gross national income, another way of removing price distortions.

The point is that endless repetition of false statements does not make them true, no matter how often one hears those claims.

Monday, April 29, 2019

Full 5G Coverage Not Until 2028, or Later, in Some Markets

Service provider executives surveyed by 451 Research believe 5G deployment will occur by 2021 or 2022 at the latest, with full network coverage happening by 2028 or even later.

The majority of respondents said they will not achieve total 5G coverage until 2028 or later. North America is expected to have the highest percentage of early deployments, driven by AT&T, Verizon, Sprint and T-Mobile. Some 47 percent of North American respondents say they will have total 5G coverage by 2025 to 2027.

More than 78 percent of respondents from the rest of the regions, excluding the Middle East and Africa, said they don’t expect to complete 5G deployment until beyond 2028.

Those predictions are mostly in line with actual experience of earlier mobile generation network deployments.

The analysts predict deployment will occur gradually, with North America setting the pace and the rest of the world playing catch-up.


Critics are Wrong About Millimeter Wave

One hears a lot of noise, much of it uninformed, about use of millimeter wave spectrum to support 5G (and all subsequent mobile next-generation networks). A growing number of voices warn that millimeter wave spectrum is the “wrong way” to implement 5G.

Such comments have a point: all other things being equal, lower-frequency 5G signals have a coverage advantage. Wider coverage, at lower cost, will be possible when sub-1 GHz assets can be used.

But coverage and capacity always are trade offs. A 5G network using sub-1 GHz assets will not be able to provide the capacity of millimeter wave assets.

Also, millimeter wave assets are the future of all future mobile networks. The need will not be as obvious in some markets as in others, but it represents the future.

That is obvious, even when voices question its use in early deployments.

What observers often forget is that mobile networks must offer both coverage and capacity. And there are a limited number of spectrum assets available to support coverage requirements. There is, however, a nearly-infinite requirement for increasing capacity.

And that capacity can only come from use of millimeter wave assets, as the International Telecommunications Union, even when urging circumspection about where 5G will make sense, attests.

Back in 2018, the International Telecommunications Union noted “concern that 5G is premature.” A report indicated “operators are sceptical about the commercial case” and  “high-levels of investment.”

But the report also contains the rationale for 5G. Simply, a new mobile next-generation network is introduced about every decade. That does not mean it gets introduced everywhere, nor at the same time. But nothing is going to stop 5G from being deployed, simply because network capacity demand drives it.

New use cases always are a feature of each touted digital network, but the practical reality is that, of the two main tools for increasing capacity--smaller cells and new spectrum--each next-generation network has brought additional spectrum resources.

Also, despite much concern that millimeter wave communications will be too hard to deploy commercially, even the ITU has noted that 5G--and all subsequent networks--will have to use millimeter wave assets.

Though many spectrum bands will support 5G--lower frequency including 600 MHz or 700 MHz, mid-band assets in the 3 GHz to 4 GHz region and millimeter wave region--capacity needs, even after spectrum refarming, will leave only the millimeter bands available for most of the future capacity supply.


Much of the concern about millimeter wave assets, and the supposed “path to 5G,” is misplaced. Millimeter wave assets will anchor every generation of mobile network from now on, for capacity. There are a limited number of coverage bands, and those bands will not be able to support capacity requirements easily.

Some of us might draw the opposite conclusions others do related to the wisdom of early user of millimeter wave technology and platforms. Specifically, millimeter wave is the future. Also, millimeter wave is the best way to supply needed capacity, going forward.

Some worry Verizon and AT&T err in emphasizing millimeter wave. On the contrary, some of us might argue, it is a tool those two carriers must use, as a practical matter. With most of the subscribers, those two firms have the greatest needs for additional capacity and spectrum.

Longer term, both will acquire operating experience that might well prove useful in the future, since all mobile generations from now on will have to rely on millimeter wave for capacity requirements.

Sunday, April 28, 2019

Could AT&T FTTH Footprint Reach 44 Percent by End of 2019?

What does it mean that AT&T will have 14 million fiber to the home passings by about the end of 2019? In a broad sense, FTTH means a chance for AT&T to retake market share from cable TV operators, which have about 65 percent of the installed base of total U.S. consumer internet access connections.

“Whenever we go into a neighborhood and turn up fiber, 25 percent (take rate) comes fast and 50 percent is eminently achievable,” said Randall Stephenson, AT&T CEO. “And we actually think we can hopefully get beyond 50 percent as we continue to get this build completed.”

AT&T’s fixed network could represent--on the high side--perhaps 62 million consumer locations passed. That figure has to be interpreted. It could mean physical locations passed. It could mean dwelling units reached.

My own understanding is that this figure refers to dwelling units, not buildings. Here’s the difference: the U.S. housing stock is divided between detached houses and multiple dwelling units and other types of housing.

In 2000, detached housing represented about 60 percent of all U.S. housing units, according to the U.S. Census Bureau. In 2017, detached homes represented nearly 62 percent of housing.

So if AT&T’s fixed network is the same as the national average, AT&T’s network might pass 37.2 million single family homes. The rest of the housing units are apartments, condominiums, other forms of attached housing, mobile homes, boats and trailers.

So assume there are 24.6 million attached dwelling units in AT&T’s fixed network footprint. One has to estimate “locations” to be served from the dwelling unit counts. We will exclude boats, trailers or mobile homes as feasible FTTH locations. Assume that “locations” (buildings) represent about 28 percent of dwellings (by definition, an MDU is one location with multiple dwellings).

In that case, there might be some 6.9 million MDU locations in the AT&T fixed network service territory. That blends MDUs of all sizes into a composite average of 3.5 units per building. So make the universe of residential locations 31.5 million.

AT&T says it will have 14 million FTTH locations in service by the end of 2019. Assuming 100 percent of those locations are single-family homes, AT&T FTTH locations would be about 44 percent, with most of the rest served by fiber to the neighborhood. It is unclear how many all-copper lines remain in service, but it is possible there are as few as a million.

Still, AT&T’s interest in high-capacity access that costs less than FTTH remains. For even if it were able to boost its market share (installed base) of consumer internet access to as much as 50 percent, half the assets would still be stranded, producing no revenue.

For AT&T no less than Comcast, lower-cost infrastructure provides two benefits: fewer stranded assets and a lower-cost base, which provides more room to either lower retail prices or boost profit margins.

Keep in mind that 25 percent take rates also imply 75 percent stranded assets. Fixed wireless built on the 5G mobile network has clear potential advantages, including lower incremental cost to supply the equivalent of fixed access and lower total capex, with lower stranded asset risk.

Saturday, April 27, 2019

Will 5G be Adopted Faster than 4G?

Will 5G subscriptions grow slower than 4G, faster or at the same rate? It is anybody’s guess right now. Researchers at CCS Insight believe 5G subscriptions will grow slightly faster than 4G. Others believe the uptake will be noticeably slower.

But even the CCS estimate anticipates that the number of 5G mobile connections will surpass one billion in 2022, taking a year less than it took 4G connections to pass the same number.

If I had to guess right now, I’d err on the side of slower uptake of 5G, for several reasons. Handset prices now are higher than when 4G first launched, so that is slowing handset replacement cycles.


Also, the performance advantage of 5G, though perhaps as much as an order of magnitude over 4G, does not arguably enable new apps that most users will want to engage with. In part, that also is because 4G speeds are getting fast enough to support streaming video easily, and that seems to be the one consumer app that really drives bandwidth consumption and is most affected by slow speeds.

But many users will find 4G getting so much better than 5G will not actually produce a noticeable difference. In other words, the app value or use case enablers are not so clear, for most users.

It actually is not clear to some of us that mobile 5G actually provides noticeable advantages in experience for users in the markets where it is being deployed. It is a bit like gigabit fixed network speeds. It is a bragging right and is “nice to have,”  but does not actually enable any new consumer app anybody can point to.

For mobile 5G, that might well be the case as well. The one area where 5G really could produce a new use case is fixed 5G, as a substitute for a fixed internet access connection. That might also be true for some mobile connections as well.

“Right now from a 5G standpoint, what we're seeing in terms of adoption tends to be business,’ said Randall Stephenson, AT&T CEO. “In fact it's exclusively business for us right now.”

In keeping with the mobile substitution framework, AT&T finds early business buyers often are using 5G as a substitute for fixed network access.

“It’s serving as a land replacement product,” said Stephenson.

Thursday, April 25, 2019

Digital One Rate Changed the Industry

Few now recall it, but AT&T Wireless once drove rapid mobile adoption by consumers by revolutionizing the way consumers paid for--and how much they paid for--long distance calling.

Back in 1998, AT&T unveiled a major new pricing plan for mobile users that priced all domestic calls for 10 cents a minute. That Digital One Rate plan effectively erased the distinction between local and long distance calling and provided a major incentive for consumers to buy mobile service.

Even many astute industry watchers did not appreciate the revolutionary nature of the plan. Technology pundit Walt Mossberg once called Digital One Rate “marketing hype.”

“It promises to make your cell phone as simple and affordable to use as a landline phone, so that you'll use it even for casual calls without a second thought,” said Mossberg. “The actual service behind the marketing isn't good enough to really allow that. I've been frustrated with it again and again.”

Digital One Rate was anything but hype. It really did revolutionize the way consumers and businesses made calls. It really did begin the mass adoption of mobile phones and the demise of landline service.

Beyond that, Digital One Rate changed the key profit drivers in telecom, from long distance revenue to mobile service. Within a decade, the industry profit driver changed from long distance to mobility. The shift from voice to data would follow.

In the ensuing years after Digital One Rate was introduced, long distance revenue plummeted, as did use of landline phone services. Digital One Rate was the driver, as the whole industry shifted pricing and packaging.


Could such a shift occur again, in a new context?

One of the obvious commercial ramifications of virtualized networks featuring network slicing is the ability to create customized features related to bandwidth, latency, security and quality of service. Prices and service levels for mobile internet access should be an early place to look for evidence of such customization.

At the most basic of levels, standard practices in fixed networks, such as the ability to buy faster or slower services, with bigger or smaller usage allowances, are commonplace. That has not been possible in the mobile segment of the business, but will be. We refer here not to the setting of tariffs for 5G than might be higher than 4G, for the same amount of usage.

What likely will be new is the ability to create speed and usage tiers on an account-by-account basis. That is new. But so far, the leading mobile service providers announcing 5G service have chosen not to introduce such plans.

To be sure, Verizon charges about $10 a month extra for data plans with comparable 4G usage allowances. And since 5G is faster than 4G, there is an effective price premium for 5G. What Verizon has not done is create speed tiers for mobile access that resemble fixed network access.

In other words, customers cannot buy 5G plans with “speeds up to 150 Mbps” as one plan, and “speeds up to 300 Mbps” as a different plan, as is the norm for fixed network access.

But AT&T might choose to do so, at least eventually. So, too, will Verizon, in all likelihood.

“I will be very surprised, if as we move into wireless, the pricing regime and wireless doesn’t look something like the pricing regime you see in fixed line,” said Randall Stephenson, AT&T CEO.

“If you can offer a gig speed, there are some customers that are willing to pay a premium for 500 meg to a gig speed and so forth,” he said. “So I expect that to be the case.”

But it will not happen right away. “We are two or three years away from seeing that play out,” said Stephenson.

Network slicing most often is thought of in the context of customized virtual networks for enterprise customers. It might also be the way mobile operators create, for the first time, speed tiers for mobile internet access that are differentiated by speed, not just usage allowance.

Whether other quality of service metrics also become standard features of some plans or not--for consumers--remains to be seen.

But this is a big potential change, and will likely create higher revenue potential for the firms that offer such speed-differentiated or possible QoS-based service plans.

Perhaps that will not be as revolutionary as Digital One Rate was. But few innovations are.

Tuesday, April 23, 2019

Give Verizon Credit for Knowing What it Has to Do in Terms of 5G Capacity and Coverage

Sometimes we do not give Verizon enough credit for knowing precisely what it has to do, capacity-wise, and for having a rational strategy to deal with those challenges. Observers have noted for years that Verizon “needs more spectrum.”

Verizon’s answer has been to use millimeter wave spectrum, small cells and spectrum sharing, in addition to acquiring new millimeter wave assets and refarming 3G spectrum. There is an unsaid ability to possibly augment with other assets as well, but Verizon is moving ahead on the assumption it must do so without spectrum asset acquisitions from other service providers.

“One of the most important features that I have talked about is of course the dynamic spectrum sharing that will come during next year, where you basically also can use, deploy wherever you are with 5G and then you don't need to allocate certain spectrum to certain technology,” said Hans Vestberg, Verizon CEO.

An important point is that there are a growing number of ways to increase effective bandwidth beyond buying spectrum licenses. Small cells, dense fiber networks, spectrum sharing and spectrum aggregation all can expand capacity.

That said, both AT&T and Verizon have committed to lots of new spectrum in the millimeter wave regions. Some have criticized such moves, arguing that volume production of radios and devices eventually be strong in the mid-band spectrum areas. There is some merit to such arguments.

But it also is true that future 5G capacity growth will have to come from the millimeter wave region, and that this will be a global trend, not something mostly limited to the United States.

Verizon realizes millimeter wave spectrum “is not the coverage spectrum,” Vestberg said. That is one reason why “spectrum sharing will be the next step for us to see that we have all the assets to deploy our strategy on 5G to meet the different type of use cases.”

No 5G service provider in the mobile business can escape an iron law of bandwidth, namely that lower frequency spectrum is better for coverage, while higher frequency spectrum is better for capacity, and mobile service providers need both.

That means operators always must balance coverage and capacity. But there are lots of moving parts. End user demand always changes, but most of the demand for capacity happens when users of smartphones are stationary, at home or at work. So most of the demand for capacity happens at home and at work.

“Remember, the majority of all the traffic is in dense urban areas, where we are now initially are focusing” its millimeter wave deployments, said Vestberg.

The point is that observers sometimes do not give Verizon strategists enough credit for having thought through capacity expansion alternatives and approaches.

New use cases might require significant amounts of additional bandwidth when users are fully mobile, but that will occur over time. For the most part, mobile bandwidth has to grow, but not at the same rate as “tethered but not moving” bandwidth demand. And that appears to be where spectrum farming and spectrum sharing will be crucial.

If Sprint and T-Mobile US do not talk as much about that, at the moment, it is because they face different challenges and own different assets.

AT&T and Verizon have heavily-loaded networks and require more bandwidth because they have more customers.

T-Mobile US and Sprint networks are relatively lightly loaded, and therefore can get by with less incremental bandwidth.

Service providers also tend to own distinct blocks of spectrum. AT&T and Verizon have more lower-band spectrum than T-Mobile US and Sprint, but that spectrum also is heavily loaded. T-Mobile has more flexibility in that area. Sprint has lots of mid-band spectrum, but not enough customers to justify aggressive deployment of those resources at the moment.

Some have noted that Verizon has less capacity per account than does AT&T, and “needs” more mid-band spectrum. Verizon technologists have run the numbers and concluded that small cell architectures--always a practical way to expand bandwidth--will do much of the job. Millimeter wave spectrum and spectrum sharing will help: the former with capacity needs, the latter for coverage.

Beyond that, there are additional mid-band resources potentially available. Dish Network’s spectrum remains a wild card. It always is possible that a T-Mobile US merger with Sprint is approved with significant spectrum sales. That remains a possible source of additional Verizon spectrum.

But none of that is essential. Every 5G service provider eventually has to supply both coverage and capacity. Capacity has to come in different ways than coverage. Verizon knows that.

Millimeter Wave Performance Will Keep Getting Better

It would be fair to say that U.S. mobile operators deploying millimeter wave spectrum remain in a learning mode. So far, it seems there is evidence of signal attenuation worse than expected as well as better than expected.

T-Mobile US, which has significant low-band assets with which to deploy 5G, argues that millimeter wave will never be used in rural areas for 5G support. But we are just at the beginning of commercial use of millimeter waves for mobile communications.

Historically, one would have been more accurate being an optimist than a pessimist about microwave signal propagation, it is fair to say. And we will have to use millimeter waves, to increase the amount of bandwidth mobile networks can supply.

Ignoring the vested interests and assets each mobile service provider possesses and touts,  many note there is a lack of available U.S. lower-band spectrum to devote to 5G.

In this illustration by Anokiwave, the orange blocks represent the 11 GigaHertz of new spectrum to be allocated by the U.S. Federal Communications Commission, and most nations eventually will do something similar. The big takeaway is how much more new spectrum will be available, compared to all mobile spectrum available up to 4G.

The skinny blue bands to the far left show currently-allocated mobile spectrum. Basically, more than an order of magnitude (10 times) more mobile bandwidth is being commercialized. And that new bandwidth is more efficient, in terms of bits per Hertz, than in the sub-3 GHz bands by perhaps three times, based on antenna techniques, alone. Modulation also makes a difference, but sheer bandwidth also matters.

The Nyquist rate  


The red and green blocks show spectrum already allocated to defense entities, for satellite or aerospace purposes.

Even as some argue 5G will be spectrum agnostic and can use “all frequencies,” some insist there is a real disadvantage to using millimeter "versus" lower-frequency spectrum. Ignore for the moment the obvious marketing stances taken by firms with different assets. Firms will tout what they have.

Some larger realities are not changed by 5G. There always is a trade off between capacity and coverage, when using wireless spectrum at any frequency. Networks always can get more-extensive coverage using lower frequencies, at the expense of potential bandwidth. Conversely, they always can get greater capacity, at the expense of lesser reach, with higher frequency spectrum.

And since “everyone” agrees a wide variety of low, medium and high frequency spectrum will be used by 5G operators, it makes sense to figure out what use cases (beyond coverage and capacity in general) are best suited to which frequency bands.

The area of greatest discovery will be in the millimeter wave bands, which have never been commercially viable for mobile services in the past. Millimeter wave spectrum will enable the greatest changes in capacity (bandwidth), and likely will be the area where the greatest number of new use cases will be developed.

Generally speaking, the gigabit speeds touted by 5G backers will be possible only when using millimeter wave assets. Generally speaking, lower-frequency spectrum will offer speeds higher than 4G, but perhaps not typically more than twice as fast.

That might suggest it is fruitful to look at 5G using lower or mid-band spectrum as important for supplying faster mobile internet for smartphone users, but not, in itself, a huge driver of new applications. One might argue that the higher bandwidth will make visual apps more compelling, and that is correct, so long as tariffs allow the higher usage.

In other words, much-faster speeds will not lead to as much innovation unless tariffs for usage are low enough that barriers to usage are not created. As one example, 5G might enable mobile TV. But consumers will constrain usage unless tariffs and usage plans encourage--or at least do not discourage--such usage.

Different use cases beyond “faster consumer smartphone access”  will tend to require millimeter wave. It will be hard to make a decent business case for fixed wireless--as a substitute for fixed network internet access--without using millimeter wave assets. There simply is no way to replicate the low cost per bit of fixed network services without millimeter wave capacity gains.

Latency performance should not vary between services at any frequency, but will be affected by the presence or absence of infrastructure edge computing facilities. Enterprises using their own edge computing platforms and private 5G will not generally have to worry about latency.

Visual applications and real-time control operations likewise are areas where millimeter wave should have relevance, often in conjunction with edge computing to control overall latency when analytics must be applied.

What sometimes is overlooked in discussions of “millimeter versus lower-band spectrum” is that many new use cases depend as much on edge computing as they do on bandwidth or latency. Millimeter wave communications will have a huge advantage where bandwidth intensity is high, but also will tend to be use cases where local processing also is necessary, hence edge computing will be necessary.

And many use cases said to be candidates for millimeter wave and edge computing (virtual reality, augmented reality, visual health applications) also will occur either indoors or in stationary or low-speed settings ideally suited to private 5G or small cell supported public networks, with local processing (edge computing).

The point is that many of the brand-new use cases (beyond faster mobile internet) will happen in settings where millimeter wave coverage is not a great issue. Most usage happens in a small number of cell sites for any network (2G to 5G). Those cell sites always are in urban areas. So 5G “coverage” is a bit of misplaced concern.

Even when wide area 5G coverage is supported, the really-high bandwidth features (gigabit or higher) will not be supported everywhere. The trade offs of coverage versus capacity remain.

So the place to look for truly-new use cases is in areas other than “faster smartphone access.” It is the urban places, indoors and outdoors, where millimeter wave and edge computing exist, where the new use cases and revenue opportunities will emerge.

The point is that ubiquitous 5G coverage is not necessarily required for the development of most new 5G-enabled apps and use cases. Those are going to happen in the dense urban areas--or inside enterprise facilities--where most economic activity takes place.

"Tokens" are the New "FLOPS," "MIPS" or "Gbps"

Modern computing has some virtually-universal reference metrics. For Gemini 1.5 and other large language models, tokens are a basic measure...