AT&T’s Broadband Trials

John Donovan, the chief strategy officer for AT&T, spoke at the Mobile World Congress recently and said that the company was trying five different technologies for the last mile. This includes WLL (wireless local loop), G.Fast, 5G, AirGig and fiber-to-the-premise. He said the company would be examining the economics of all of different technologies. Let me look at each one, in relation to AT&T.

Wireless Local Loop (WLL). The technology uses the companies LTE bandwidth but utilizes a point-to-multipoint network configuration. By using a small dish on the house to receive the signal the company is getting better bandwidth than can be received from normal broadcast cellular. The company has been doing trials on various different versions of the technology for many years. But there are a few recent trials of the newest technology that AT&T will be using for much of its deployment in rural America as part of the CAF II plan. That plan requires the ISP to deliver at least 10/1 Mbps. AT&T says that the technology is delivering speeds of 15 to 25 Mbps. The company says that even at the edge of a cellular network that a customer can get 10 Mbps about 90% of the time.

G.Fast. This is a technology that uses high frequencies to put more bandwidth on telephone copper wire. Speeds are reported to be as high as 500 Mbps, but only for very short distances under 200 feet. AT&T recently announced a G.Fast trial in an apartment building in Minneapolis. The technology is also being tested by CenturyLink and Windstream. All of these trials are using existing telephone copper inside of existing apartment buildings to deliver broadband. So this is not really a last mile technology. AT&T brings fiber to the apartment complex and then uses G.Fast as an inside wire technology. If they find it to be reliable this would be a great alternative to rewiring apartments with fiber.

5G. AT&T recently announced a few trials of early 5G technologies in Austin. They are looking at several technology ideas such carrier aggregation (combining many frequencies). But these are just trials, and AT&T is one of the companies helping to test pre-5G ideas as part of the worldwide effort to define the 5G specifications. These are not tests of market-ready technologies, but are instead field trials for various concepts needed to make 5G work. There is no doubt that AT&T will eventually replace LTE wireless with 5G wireless, but that transition is still many years in the future. The company is claiming to be testing 5G for the press release benefits – but these are not tests of a viable last mile technology – just tests that are moving lab concepts to early field trials.

AirGig. This one remains a mystery. AT&T says it will begin trialing the technology later this year with two power companies. There has been a little bit of clarification of the technology since the initial press release. This is not a broadband over powerline technology – it’s completely wireless and is using the open lines-of-sight on top of power poles to create a clear path for millimeter wave radios. The company has also said that they don’t know yet which wireless technology will be used to go from the poles into the home – they said the whole range of licensed spectrum is under consideration including the LTE frequencies. And if that’s the case then the AirGig is a fiber-replacement, but the delivery to homes would be about the same as WLL.

FTTP. Donovan referred to fiber-to-the-home as a trial, but by now the company understands the economics of fiber. The company keeps stretching the truth a bit about their fiber deployments. The company keeps saying that they have deployed fiber to 4 million homes, with 8 million more coming in the next three years. But the fact is they have actually only passed the 4 million homes that they can market to as is disclosed on their own web site. The twelve million home target was something that was dictated by the FCC as part of the settlement allowing the company to buy DirecTV.

We don’t know how many fiber customers AT&T has. They are mostly marketing this to apartment buildings, although there are residential customers around the country saying they have it. But they have not sold big piles of fiber connections like Verizon FiOS. This can be seen by looking at the steady drop in total AT&T data customers – 16.03 million in 2014, 15.78 million in 2015 and 15.62 million at the end of the third quarter of 2016. AT&T’s fiber is not really priced to be super-competitive, except in markets where they compete with Google Fiber. Their normal prices elsewhere on fiber are $70 for 100 Mbps, $80 for 300 Mbps and $99 for a gigabit.

Technology Predictions

Alexander_Crystal_SeerThroughout history there are examples of people publicly declaring that something wasn’t possible and then a few years later the predicted impossible happened. There are some well-known examples of this in our own industry. Consider the following:

In 1961 Tunis Craven, an FCC Commissioner said, “There is practically no chance communications space satellites will be used to provide better telephone, telegraph, television, or radio service inside the United States.” Four years later the first commercial communications satellite was launched.

In 1878 Sir William Preece, the chief engineer of the British Post Office declared, “The Americans have need of the telephone, but we do not. We have plenty of messenger boys.”

The same year an internal memo at Western Union opined that, “This ‘telephone’ has too many shortcomings to be seriously considered as a means of communication. The device is inherently of no value to us.”

The famous movie producer Darryl Zanuck was quoted in 1946 as saying, “Television won’t be able to hold on to any market it captures after the first six months. People will soon get tired of staring at a plywood box every night.”

In possibly the most famously wrong technology quote Bill Gates said, “We will never make a 32 bit operating system.”

Along the same lines, Ken Olson, the President of Digital Equipment Corporation and a major manufacturer of mainframes said in 1977, “There is no reason anyone would want a computer in their home.

Our industry is full of predictions about the future. You always have to wonder which of these predictions will come true and which will prove to be totally wrong. It seems predictions are of two types. Some predictions are made about working technologies that aren’t going to make a dent in the marketplace. Our technology history is full of devices that nobody wanted to buy. But there are also many predictions made, like the examples above, about the limitations of technology and what is possible.

I know that technology improvements have taken me by surprise a few times. If in the early 2000s you would have asked me to predict the top data speeds that could be achieved on telephone copper or on coax I would have underestimated the speeds by a magnitude or more. At that time scientists in the labs all said that there were insurmountable interference issues that made higher frequencies unusable on both kinds of copper.

But some smart scientists doggedly worked on these problems and today we have G.fast that is putting gigabit speeds for short distances over telephone copper. And we have the potential for incredibly fast speeds on coax. I’ve witnessed a lab test that put 6 gigabits through a piece of coax.

I make a lot of predictions in this blog. But one thing I’ve learned is that you should never say never when it comes to technology (well, except maybe for tabletop fusion power). Instead predictions are better made talking about the likelihood of something being achieved in the foreseeable future versus the distant future.

A good example of this is gigabit cellular service. Unfortunately some of the press has confused millimeter wave wireless with 5G cellular and there are articles all over the web talking about the coming gigabit cellular service. Is gigabit cellular possible? I would venture to say that in a lab setting with a small number of phones this might be possible today or in the near future.

But there are physics limitations that limit gigabit wireless speeds to short distances and for this technology to ever become pervasive we would have to massively rework all cellular infrastructure to literally surround ourselves with cellular transmitters. It is that limitation that means that this is an extremely unlikely application within any reasonable time frame. It’s certainly possible someday that we might be surrounded by tiny IoT devices that can somehow work as a mesh network to bounce around fast data signals. But there are a whole lot of technology breakthroughs needed first to implement such a technology. So is gigabit wireless possible – I think it is. Will we see it in our lifetimes other than perhaps in a few controlled settings – I predict not. Guess we’ll have to wait to see if I’m right.

An Upgrade to G.fast

Speed_Street_SignNokia has announced the lab trial of the next generation of G.fast, the technology that can pump more bandwidth through telephone copper. They ae calling the technology XG.fast.

In a recent trial the equipment was able to send a 5 Gbps signal over copper for 100 meters and 8 Gbps for 30 meters. This is much faster than the G.fast top speed in trials of about 700 Mbps. In a real life situation using older copper the speeds will not be nearly this fast. G.fast in real life trials has gotten about half of the speeds seen in labs, and it would be impressive if that can also be achieved for XG.fast.

The technology works by utilizing higher-band frequencies on the copper. Traditional VDSL uses frequencies up to about 17 MHz. G.fast uses frequencies between 106 MHz and 212 MHz. XG.fast climbs the spectrum even further and adds on spectrum between 350 MHz and 500 MHz.

There are a lot of issues involved in using all of this frequency on a small-gauge copper. The main problem is crosstalk interference – when adjoining copper wires interfere with each other, and this degrades the signal and drastically cuts down on the distance the signal can be transmitted.

Nokia mitigates the crosstalk using vectoring, the same as is done with VDSL and other DSL technologies. Vectoring generates an –out of-phase signal that can cancel out some of the interference. But there is so much interference at thise frequencies that vectoring can only keep the signal coherent for the short distances seen the trial.

To date there has not been a lot of interest in G.fast. Adtran, the other competitor in the G.fast space claims to have now conducted ninety field trials of the technology worldwide. That’s an extraordinarily low number for a technology that can add speed to existing copper. But it looks like most phone companies are not interested in the technology, and they have some good reasons.

The short distances make G.fast and its new successor impractically expensive in the copper plant. In order to use the technology the telco would have to mount an XG.Fast transmitter at the pole outside each home, or in dense neighborhoods to perhaps serve a few homes. But if the telco wants to take advantage of the faster speeds that XG.Fast can get into the home they also would need to string fiber to feed the XG.Fast transmitters.

XG.Fast is largely a fiber-to-the-curb technology and the cost of the building fiber up and down streets is the big hurdle to using the technology. Any company willing to spend the money to build that much fiber probably isn’t willing to trust copper for the last 100 feet.

There is one application where XG.fast makes good economic sense. It can be extremely costly to rewire older apartment buildings with fiber. But every apartment building has existing telephone wiring and XG.fast can be used to move data from a telephone closet to the apartment units. This sounds to be far less costly than trying to snake fiber through older buildings. Since a lot of companies have avoided older apartment buildings this might offer a relatively inexpensive way to bring broadband.

You can’t fault Nokia for continuing to pursue the technology. There is a huge amount of copper still hanging on poles and the world keeps shouting for more broadband. But I get nervous about recommending any technology that isn’t widely accepted. I can picture a telco deploying this technology and then seeing support dropped for the product line.

But I can’t see this ever being much more than a niche technology. Telcos in the US seem to be looking for reasons to tear down copper and don’t seem willing to take one more shot at a copper technology. There might be a good business case for using the technology to extend broadband inside older buildings. But US telcos seem completely uninterested in using this in older copper networks.

New Technology – May 2016

light beamsThe following are some new breakthroughs that might eventually benefit our industry:

Better Long-haul Fiber. Researchers at the Moscow University of Physic and Technology, ITMO University of St. Petersburg, and the Australian National University have demonstrated a new technology that can drastically increase the efficiency of long-haul fiber.

They have found that introducing silicon nanoparticles into a fiber optics path can increase what is called the Raman effect. The Raman effect is where light interacts with some materials to produce wavelengths of different colors. In these materials the light causes the affected molecules to increase in energy, and at the higher energy level the materials then re-emit a photon that has a lower energy level than the original light stream.

Today’s lasers use metallic particles to induce the desired color of wavelength but the silicon nanoparticles generate light nearly 100 times stronger than the technology used today. While it will take a while to go from lab to production, this has a huge potential for the efficiency and the distance between repeaters on long-haul fibers.

A New Form of Light. Scientists at Trinity College in Dublin Ireland have been able to produce a new kind of light.  Physics has viewed the properties of light as a fixed constant. This was based upon the angular momentum of photons and Planck’s constant. But the scientists have been able to produce a form of light with an angular momentum that is half the value of Planck’s constant.

Scientists have long theorized that different fractional angular momentum is possible but this is the first time it’s been produced. The first potential use for this new form of light is with fiber optics transmissions. This new kind of light looks to have properties that would allow for the transmission of significantly more bits of data than with normal light.

Faster G.Fast. Israeli chip-maker Sckipio has developed a G.Fast chipset that will double the effective speed of G.Fast. The chip can support symmetrical throughput speeds of 750 Mbps. Sckipio says that they already have another chip set on the drawing board that might double that speed to about 1.5 Gbps.

Their chipset is the first G.Fast design to have fast speeds in both directions and provides greater overall data throughput. While the only trial of G.Fast in the country that I’ve heard about is being done by CenturyLink, it’s been reported that AT&T is thinking about adopting the technology. The company has made numerous announcements about expanding their U-verse product to millions of homes and G.Fast is basically a fiber-to-the curb product that would let them string fiber in neighborhoods but use the existing copper network to bring the bandwidth into the home.

New Data Storage Technology. IBM Research announced the first successful trial of storing data using phase-change memory (PCM). This technology can store 3-bits of data instead of just one.  There are many advantages of PCM storage – it retains memory without power, it allows for faster read / write and it can be overwritten over 10 million times (compared to flash drives which wear out after around 3,000 write cycles).

IBM sees PCM being used in conjunction with flash storage to allow for very fast launching of devices like cell-phones and computers. It would also allow for much faster inquiries, speeding up computer processing speeds on any device.

 

US Telcos Indifferent to G.Fast

Speed_Street_SignG.Fast is a new technology that can deliver a large swath of broadband over copper wires for a short distance. The technology uses some of the very high frequencies that can travel over copper, much in the same way that DSL does for lower frequencies.

The International Telecommunication Union (ITU) just approved the final standard for the technology with the G.9701 standard for “Fast Access to Subscriber Terminals.” Several vendors including Alcatel-Lucent and Huawei have been producing and testing units in various field trials.

British Telecom has done a number of these tests. The largest such test was started in August for 2,000 customers in Huntingdon, Cambridgeshire. During the trial they are offering customers speeds of 330 Mbps, but they expect at the end of the trial to be able to raise this to about 500 Mbps.

The technology involves building fiber along streets and then using the existing copper drops to bring the bandwidth into the home. This is the most affordable kind of fiber construction because a telco can overlash fiber onto its existing copper wires on the poles. That means very little make-ready work, no permits needed, and no impediments to quick construction. This kind of fiber construction can literally be done at half of the cost faced by other fiber overbuilders.

British Telecom has done a number of trials across the country. Alcatel-Lucent has also done trials with Telkom Austria. But for the most part American telcos have shown no interest in the technology. The only real trial here that I’ve read about is a trial with CenturyLink in Las Vegas.

And I frankly don’t understand the reluctance. G.Fast is a halfway solution on the way to a full fiber deployment. As cable companies and overbuilders like Google are stepping up deployment of gigabit speeds, either through fiber or through fast cable modems using DOCSIS 3.1, the telcos have been announcing fiber builds to remain competitive. AT&T has announced gigabit fiber builds in more than twenty markets. CenturyLink says it will be passing 700,000 homes with fiber in 2016.

So why wouldn’t an American telo seriously consider G.Fast? With capabilities up to 500 Mbps in real-world applications it gives them a product that can compete well with other fast technologies. And by overlashing the fiber to deploy G.Fast the telco will have tackled one of the major costs of building an FTTP network, by getting the fiber deep into the network. And with G.Fast a telco can avoid the expensive fiber drops and electronics which are the most expensive part of a FTTP network for them.

I could envision somebody like CenturyLink building fiber to the more lucrative parts of town while deploying G.Fast to older copper neighborhoods. This would give them a far greater fast broadband coverage, making it easier and more cost effective to advertise their broadband.

But it seems like most of the US telcos just want out of the copper business. And so, rather than take this as an opportunity to milk another decade out of their copper networks before finally building fiber, they seem prepared to cede even more broadband customers to the cable companies. That has me scratching my head. The cable companies have clearly accepted that their entire future is as ISPs and that data is the only real product that will matter in the future. It just seems that the large telcos have not quite yet come to this same conclusion.

Large Telcos and DSL

Copper wireThere has been a spate of articles recently talking about how the number of cable customers at the large cable companies took their first big dip last quarter. This was the first time when the cable industry as a whole saw an overall significant customer loss, and this raises the question the question if cord cutting is real.

But there was another significant statistic in these same press releases. AT&T and Verizon together lost 474,000 DSL customers in the second quarter of 2015. The two made up some of these losses by adding 313,000 data customers to their FiOS and U-verse networks, so certainly some of the losses are offset by customers who shifted from DSL to something faster.

But this continues the trend that these two largest telcos are shedding DSL customers. The numbers just keep growing and this is the first time that number approached half a million customers.

Verizon has made it clear for years that they have no love for their copper networks. They have been selling significant chunks of the older networks to Frontier. They have been pestering the FCC for years to be able to turn down the copper in neighborhoods where they already have FiOS fiber.

Perhaps more surprising is that Verizon recently sold a significant number of FiOS customers to Frontier, and I have speculated before that Verizon doesn’t want to stay in any landline business. When you read their annual reports, any mention of their landline business is buried deep inside and they obviously have put all of their emphasis on the wireless business.

AT&T is a bit more perplexing. They have not been selling copper customers. But they have told the FCC a number of times that they would like to walk away from millions of customers on rural copper networks. AT&T recently promised the FCC as part of the DirectTV deal that they would aggressively add new broadband customers. While they have insinuated to the FCC that the new customers would all be on fiber, I would not be surprised to see a lot of them on U-Verse.

Many people speculate why AT&T bought DirectTV. My guess is that they want to get out of the business of delivering video over wires. U-Verse becomes a much better data product  if it doesn’t have to carry video so that all of the bandwidth would be used for data. There must already be a lot of current U-Verse customers bumping up against their bandwidth and wanting faster data connections.

It’s also interesting that AT&T hasn’t divested of rural copper networks in the same manner as Verizon. Again, I am only speculating, but my guess is that they don’t want those networks to be revitalized and then compete against their wireless networks. I think AT&T has a long term plan to serve rural areas with wireless only.

The one shame about cutting down the copper networks, particularly in urban and suburban neighborhoods, is that those networks could be upgraded relatively inexpensively with G.fast to deliver much faster speeds. CenturyLink just announced that they are testing 100 Mbps copper in Salt Lake City. Some of the copper networks in Europe are doing this with even faster speeds and the technology is generally referred to there as fiber-to-the-curb.

But obviously both companies have decided that G.fast is not a technological path they want to follow, and both are going to be aggressively decommissioning copper over the next five years.

I don’t feel too bad about a customer who is told they have to move from a copper network to a FIOS fiber network. But I am really worried about rural customers if somebody cuts down the only telecommunications wire to their home when the copper comes down. At that point those folks are going to be paying cellphone prices for both voice and data, and for some millions of them there is not enough coverage to provide those services over cellular. I predict we are going to be cutting customers off from communications and moving parts of the country back seventy-five years. I hope I am wrong.