Wednesday, February 28, 2018

How Big a Revenue Opportunity are "Smart Cities" IoT Services?

By some estimates smart cities represent the biggest opportunity for mobile service providers.

Others might disagree, connected home or connected car or manufacturing will be bigger market opportunities.

IoT technology revenues across 12 key smart city technologies and verticals is predicted to grow from around $25 billion in 2017 to $62 billion in 2026, according to ABI Research.

Among the drivers are smart meters and video surveillance, representing $20 billion of the expected revenue.

Of course, “smart cities” might include so many verticals that it is not a single market.



Will 5G Make Packet QoS (Prioritization) Unnecessary?

Network neutrality remains as obscure a topic as ever, despite efforts to reduce it to a bumper sticker slogan. Virtually everyone agrees on the original and basic version: no blocking websites; no censoring online content; no throttling, degrading or discrimination of network performance based on content.

Also, disclosure to consumers about how networks are managed.

Where there is disagreement is when network neutrality is stretched to cover any number of other practices, such as zero rating or packet prioritization. The former is exclusively a matter of business practices. Can, or should, an internet service provider allow consumers access to services, apps or sites without charge?

The latter is a matter of importance for some classes of apps that actually require deterministic performance.

“AT&T is not interested in creating fast lanes and slow lanes on anyone’s internet,” says Bob Quinn, AT&T senior EVP. “What we do care about is enabling innovative new technologies like autonomous cars, remote surgery, enhanced first responder communications and virtual reality services, which are real-time interactive services that require end-to-end management in order to make those services work for consumers and public safety.”

“I think we can all agree that the packets directing autonomous cars, robotic surgeries or public safety communications must not drop,” said Quinn.

The Restoring Internet Freedom decision really is about freedom, no matter whether one is for or against the order. A key issue is “who gets the freedom.” Freedom is for consumers, we agree. Freedom is for app and content providers, we agree. The FCC’s Restoring Internet Freedom order adds freedom for access providers.  

If you read the document, you are quickly “in the weeds” of telecom policy. But like the decision or not, the decision is fundamentally about freedom.

At one important level, the heart of the matter is our understanding of network neutrality rules, which historically has had a weak and strong form. Though many proponents of net neutrality have argued “net neutrality is dead,” others would argue that all the FCC has done is move back to the original form of net neutrality, and away from the “strong” form some have advocated.

Paradoxically, the original “weak” form protects and enhances freedom, while the “strong” form restricts freedom.

That original form was the the Four Freedoms principles articulated in 2004 and was then formally adopted by a unanimous Commission in 2005.

These include the freedoms for consumers to (1) ‘‘access the lawful internet content of their choice’’; (2) ‘‘run applications and use services of their choice, subject to the needs of law enforcement’’; (3) ‘‘connect their choice of legal devices that do not harm the network’’; and (4) ‘‘enjoy competition among network providers, application and service providers, and content providers.’’

Repealing common carrier regulation does not repeal those principles.  “We reaffirm and honor this longstanding, bipartisan commitment,” the FCC says.

To be clear, those principles mean ISPs cannot block lawful content. The principles also have been understood to include a bar on “throttling” or degrading access to some apps based on the ownership of those apps.

Some proponents of strong forms of net neutrality rightly point out that some instances of content or site blocking by ISPs have occurred. It also is fair to say that the internet freedoms principles bar such action, and that the FCC remains committed to the “no blocking” rules.

Perhaps the bigger issue, though, is whether “best effort” access is the only class of service any internet user can buy or use. This is more complicated.

Major applications providers routinely use content delivery networks (CDNs) that improve quality of experience by reducing packet latency. In other words, CDNs do not “treat all bits equally” and instead use mechanisms to speed delivery and predictability of packet arrival beyond those possible using strict “best effort” mechanisms.

That normally is done by placing content stores closer to the edge of networks, thereby avoiding full transits of the public internet across wider distances.

AT&T’s point is simply that consumer internet access is not the issue. Though some believe there is demand for packet prioritization services, others think the market is so competitive that no ISP could actually make such a program work.

Think about coming 5G networks. If latency is just a few milliseconds, and bandwidth is upwards of several hundred megabits per second, up into the low gigabits per second, where is the value of packet prioritization?

In other words, at least some general purpose access networks will be so good that the market for prioritized delivery simply does not exist, for consumer internet services.

That would likely be true even for apps that benefit from low latency and more-predictable packet arrival (streaming video and audio; video conferencing; voice communications).

In other words, there might not be demand for fast lanes, because the general purpose networks will eliminate the need.

Monday, February 26, 2018

FCC Readies Spectrum Auctions in 28-GHz, 24-GHz Bands

Assuming the U.S. Congress clarifies rules on upfront payments, the U.S. Federal Communications Commission will hold spectrum auctions in the 28-GHz and then 24-GHz bands in November 2018, for 5G network use.


“It is my intention for the United States to hold an auction beginning this November of spectrum in the 28 GHz band, followed immediately thereafter by an auction of spectrum in the 24 GHz band,” said Ajit Pai, FCC chairman.


That will follow auctions of 600-MHz spectrum, opening up spectrum sharing for 150 MHz of spectrum in the 3.5-GHz band.


“I intend to propose the next steps needed to make the 3.7 to 4.2 GHz band available for commercial terrestrial use,” said Pai.


In addition to those spectrum moves, Pai also noted that the FCC Spectrum Frontiers Order opened up nearly 11 GHz of spectrum in the bands above 24 GHz for mobile use.


“We followed up by making an additional 1,700 MHz of millimeter wave spectrum in the 24 and 47 GHz bands available for terrestrial 5G wireless use,” said Pai. “Last year, we began to explore unlicensed use in the 6 GHz band.”


Pai says the FCC also is moving to remove barriers to wireline deployment, which addresses issues like easier and cheaper ways to attach equipment to utility poles, said Pai. “We’ve also updated our rules for high-speed, dedicated services by lifting rate regulation where appropriate.”


“In sum, we are creating huge incentives for the private sector to invest in the 21st-century networks used for backhaul,” he said.


All that new spectrum represents at least an  order of magnitude more spectrum than presently is available for all mobile service providers. Combined with small cell architectures and spectrum aggregation, there could be two orders of magnitude more effective spectrum available for 5G use.

Ninth Circuit Rules FTC Does Have Authority to Oversee Internet Access

With the caveat that the U.S. Ninth Circuit Court of Appeals often makes rulings that are overturned (79 percent), the Ninth Circuit Court of Appeals unanimously ruled that the Federal Trade Commission acted within its authority in bringing a claim against AT&T over its data-throttling practices.

Some will say that the decision validates the oversight the Federal Trade Commission will exercise over internet access services, as proposed by the Federal Communications Commission in its recent Restoring Internet Freedom order.

Window of Opportunity for Paid Prioritization is Shutting

Despite some fears, in some quarters, about paid prioritization of packets, is that a realistic fear? If the value of packet prioritization--paid or not--is guaranteed performance (low latency), 5G (and advanced 4G) should--by definition--eliminate the problem, and therefore the possibility of profiting from services that offer higher quality assurances..

In other words, the potential value of packet prioritization is quality assurance for applications that require low latency (packet arrival times).

But when a 5G network routinely has latency in single-digit milliseconds, is prioritization still needed, and if so, for what apps? In other words, does 5G actually eliminate the problem that packet prioritization is said to solve?

And, if that is true, then fears about paid prioritization are misplaced. As much as some internet service providers might like to sell packet prioritization services, there might not be a market for such services on 5G networks.

Perhaps paid prioritization could still have value on fixed networks, but user experience itself might force fixed networks to upgrade performance, simply to keep pace with 5G networks, and therefore close the window of opportunity for packet prioritization, even on many fixed networks.



The point is that the feared “packet blocking and throttling” still is barred by Federal Communications Commission policy, while “paid prioritization” has a slim window for relevance until 5G arrives.

The stated fears about removing common carrier regulation from internet access will prove misplaced.

5G Will Severely Erode or End Value of "Paid Prioritization"

Is there value in packet prioritization--paid or not--in the 5G era? The prioritized packets value proposition has been latency improvement. When a 5G network routinely has latency in single-digit milliseconds, is prioritization still needed, and if so, for what apps?

The point is that the feared “packet blocking and throttling” still is barred by Federal Communications Commission policy, while “paid prioritization” has a slim window for relevance until 5G arrives.

The stated fears about removing common carrier regulation from internet access will prove misplaced.

As often is the case, telecom regulators and industry executives do not agree on the impact of regulations.  EU Commissioner Andrus Ansip, for example, does not believe network neutrality rules hamper investment in 5G.

Ericsson CEO Börje Ekholm, on the other hand, believes such rules will prevent the creation of new 5G services. “The principle of net neutrality is not to discriminate [against], throttle or degrade based on content but not all traffic is created equally and we don't believe this will work in the 5G future," said Ekholm.  "There will be a need for a regulatory regime that allows service providers to create services that are differentiated based on user experience."

On the other hand, arguments can be made that some feared consumer internet practices, such as content blocking, app throttling or paid prioritization will emerge in the United States.

Content blocking or app throttling still are prohibited as continuing Federal Communications Commission policy, and will be enforced by the Federal Trade Commission. Paid prioritization might have little value as consumer internet speeds climb towards gigabit levels and latency drops in the advanced 4G era and 5G era towards a few milliseconds.

Under such conditions, little value is provided by a consumer paid prioritization scheme. Where, one might ask, is the business value when speeds routinely are in hundreds of megabits per second, up to gigabits per second, and latency is in single digits?

That is precisely what is coming in the 5G era.

5G-NR radio networks boosted device speeds an order of magnitude in 100 MHz of 3.5 GHz spectrum, while latency also improved by an order of magnitude (10 times), in simulations conducted by Qualcomm.

In a separate simulation using 800 MHZ of 28-GHz spectrum, browsing speeds increased by two orders of magnitude, while latency improved by two orders of magnitude.

The 3.5-GHz test, conducted in Frankfurt, increased end user speeds from 56 Mbps for the median 4G user to more than 490 Mbps for the median 5G user. Latency dropped from 116 milliseconds to 17 milliseconds.

The San Francisco simulation boosted browsing speeds from 71 Mbps for the median 4G user to 1.4 Gbps for the median 5G user.

Browsing download latency dropped  from 115 ms to 4.9 ms.

In both simulations, existing cell site locations in Frankfurt and San Francisco were used, where 5G NR cell sites are co-located with actual, existing LTE sites.

The point is that the value of packet prioritization diminishes as the routine performance of a mobile network increases.

Sunday, February 25, 2018

Historic Change in Telco Business Thinking

It is not hard to understand that telecom’s historical geographic legacy--specified areas of operation--has implications for operator efforts to grow application businesses at scale. In other words, service providers of all types are accustomed to creating services and apps that run on their own networks, in their licensed territories.

In a new ecosystem where applications run “over the top,” without regard to geography, and where revenue models often require scale, that older mentality often is unhelpful.

One salient and helpful development, in that regard, are the new OTT video streaming apps designed expressly to run on any network, at scale. We sometimes miss the importance.

Where in the past apps were designed to run on “my network,” now apps are being created to run on “anybody’s network.” In other words, the new apps are borderless (to the extent allowed by copyright rules).

That is a fundamental prerequisite for apps in the internet era, especially those with scale requirements.


This is something quite new. In the past, “geography” has been a key mental, legal and operating foundation guiding strategy, as has the notion that apps “run on our network.” In the next evolution of the industry, network footprint will be part of the business, but not its future.

As operators are shifting from “video that runs on our network and is sold to our customers” to “video that any consumer can buy from us.”



In other words, the future of apps and revenue are separated from the network, perhaps less so for mobile than fixed networks, but separated nevertheless.

So strategy will change in fundamental ways. For the first time, at scale, at least some service providers will try to create business models based on apps that run at scale on anybody’s network, the same strategy as used by Google, Facebook and other app providers.

What Will 5G Cost?

Will 5G cost so much more than 4G that the business model breaks? If orders of magnitude more bandwidth have to be delivered, and if propensity to pay does not change but incrementally, what happens to capex cost?

What happens if the number of cell sites grows two orders of magnitude?

The number of U.S. cell sites, for example, could balloon from perhaps 80,000 to as many as a million, estimates Jack Waters, Zayo Technologies CTO.

What happens to backhaul costs? Will advanced radio arrays cost less, more or the same as today's radios?

Some fear capital investment could be double to triple the levels of 4G. By other estimates, 5G will require just four percent higher capital investment than did 4G.

Those are huge uncertainties.

But cost parameters are changing so much that some expect 5G capital investment might actually be less than 4G, even if the historic trend is that each next generation mobile platform requires incrementally more investment.

Over the last several decades, infrastructure costs (transport and access) have dropped by orders of magnitude, says Waters, referring to the cost-per-bit parameters.

One reason is infrastructure cost improvements, including open source platforms that inherently cost less than proprietary solutions.

Nokia and Facebook, for example, are working on 60-GHz fixed wireless platforms for urban or suburban areas. Nokia expects to develop Facebook’s Terragraph network and will conduct trials of the technology this year.

Combining Nokia’s wireless passive optical network protocols with Terragraph's mesh-routing and multi-hop capabilities allows broadband providers to wirelessly deliver gigabit services over wider areas with high reliability and meet growing demands for ultra-broadband access, Nokia says.

Nokia and Facebook will also work together to accelerate IEEE's 802.11ay industry standard, leveraging Nokia's Wireless PON and Terragraph's TDMA scheduling capabilities, Nokia says.

Wireless PON is based on 802.11ad WiGig technology and provides a wireless drop for fiber-to-the-home networks.

Access points can be easily mounted on utility poles, street lights or a building facade, and deliver gigabit-per-second speeds to a self-installable WPON Home unit.

The point is that 5G capex requirements might not scale linearly as have prior network generations, despite the heavy reliance on small cells and need for dense optical backhaul.

In fact, unit costs for mobile data have been falling for some time.   Figure 1: The unit cost of mobile data traffic [Source: Analysys Mason, 2013]
source: Analysys Mason

Monday, February 19, 2018

90% of Internet Users Now Use Cloud-Based Apps

Perhaps 3.6 billion global consumers now use cloud computing, in the form of the apps and sites they use regularly. If there are a total four billion internet users globally, that suggests 90 percent of world internet users use cloud-based  applications and services.

That has obvious implications for the computing industry.


In 2017, Amazon Web Services generated about $18 billion of revenue for Amazon. Microsoft, which includes its cloud apps in its cloud revenue segment, booked perhaps $27.4 billion in cloud revenue.

From 2016 to 2017, AWS revenues grew 42 percent from $12 billion to $17 billion, while Microsoft's cloud revenue contributions grew about nine percent.

Looking just at customers of cloud computing services (and not including applications), AWS has perhaps 34 percent installed base; Google 20 percent; IBM 15 percent; Microsoft about 15 percent.

In 2017, enterprises spent about as much on cloud infrastructure services as they did buying servers to support their internal computing operations. But a majority of computing workload probably now happen on cloud facilities.  


More significantly, cloud spending is going to displace a greater percentage of enterprise computing spending in coming years.



Saturday, February 17, 2018

"Winning" is Not What It Used to Be

What does “winning” look like for telco internet access? In the monopoly era, this was no question at all. In the competitive era, maximum feasible market share is something else, entirely.

And that underpins nearly all business models. In the monopoly era, a network could be built on the safe assumption that upwards of 95 percent of locations passed would generate revenue.

In the competitive era, it is doubtful whether maximum possible success ever leads to market share more than 45 percent. In other words, no matter how good a service provider is, or how powerful its value proposition, more than half of all locations will not generate any revenue.

Rough implication: the cost of building a network, “per customer,” doubles.



In Singapore, SingTel, the leader, had 44 percent market share in 2014.

In Nigeria, MTN, the market share leader in 2017, had 39 percent share.

The clear implication is that no service provider, anymore, can build a network and assume it will get much more than about 40 percent to 45 percent market share, at best. In other words, more than half of capital investment in the access network will routinely be stranded.

Operating costs “per customer account” might be as much as double what they might have been if market share were closer to 95 percent.

All that constrains our notions of what “success” looks like. For no matter how much a contestant invests, it cannot reasonably expect to achieve much better than 40 percent to 45 percent market share.

That necessarily raises the danger of “over-investment” in facilities and features that will not provide an adequate financial return.

Consider a U.S. telco modeling a gigabit internet access rollout. Assume that telco has present market share of 35 percent to 40 percent. How much better can it do, if it upgrades to gigabit access?

In all too many cases, the answer is “a few points of market share.”

Consider Cincinnati Bell, which is in the process of upgrading to fiber-to-home across its service territory. “Our results demonstrate that we continue to compete and win against cable with fiber,” said Leigh Fox, Cincinnati Bell CEO.

“Competing” in this case means having 40 percent market share, and gaining about three share points in the year.

Since its main competitor Charter Communications, is in the midst of its own upgrade to gigabit speeds, the issue is how much more share Cincinnati Bell actually can take. The worst case answer might be “not much more.”

The best case answer might be 10 share points, to reach something like a 50-50 split of the market, until and unless other competitors enter the market. In that case, Charter and Cincinnati Bell might see something like a 40-40-20 or 45-45-10 share pattern, as a reasonable expectation of “winning.”

Winning is not what it used to be.

"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...