Template Tools
DSL Prime
Gigabit and More Wireless Surprisingly Close
Sunday, 06 April 2014 13:59

Stanford Professor Andrea Goldsmith believes wireless capacity can increase 50 times in the next 5 to 10 years. We'll soon have gigabits rather than the tens of megabits now typical. The cost of delivering each bit - or gigabyte - is dropping at a ferocious rate. Prices are staying high in most countries.

   Inability to sell the capacity coming online at the prices they want is far more of a problem for telcos than the wildly exaggerated spectrum "crisis." The result is a desperate effort to eliminate competitors around the world. It's easy to see price-fixing is the goal behind Sprint/T-Mobile, Bouygues/SFR, Telefonica/E-Plus, Softbank/eAccess, AT&T/Leap and the massive CEO support for ETNO's campaign to shrink the industry.

   Cell phone inventor Marty Cooper points out "we've never had a spectrum shortage." By and large, the "spectrum crisis" has been invented by politicians and lobbyists who pull politicians' strings. In all but  limited situations, there's plenty of bandwidth and spectrum.

Last Updated on Saturday, 31 May 2014 04:11
No One Buys Cable Caps at Time Warner
Thursday, 13 March 2014 15:35

MSlavik No DogFewer than 1 in 100 take 30 gig cap for $5 discount. CEO Rob Marcus reports only "thousands" of Time Warner's 11M customers are taking the discount after 6 months. He still supports the idea because he wants to charge heavier users more. "Notwithstanding the low uptake of usage-based tiers, I think it's a very important component of our overall pricing philosophy." Thanks to Jon Brodkin of Ars Technica for catching the comment and also finding a great picture. http://bit.ly/1i8rQmy Another interesting datapoint from Marcus was that only about 8% of data customers buy their own modem. His pr person was unhappy I included the modem rental in my previous articel on TWC http://bit.ly/1hfPIpR but I think that appropriate if more than 90% of customers rent.

    Marcus explained that Comcast deal was much more attractive than Charter because he had "significant concerns about the value of Charter stock" inspired me to look at the price of Charter stock. A market cap of $13B and debt of about $14B values the company at ~$6,000/subscriber. That's very hard to justify on any plausible earnings, even if John Malone is a financial magician and Tom Rutledge a strong operations manager. 

   My latest bill from Time Warner for regular cable modem service was $63/month, up about 40% in just a few years.



Last Updated on Friday, 14 March 2014 13:58
Cisco: Africa in 2017 to Have More Internet Users than U.S.
Sunday, 02 March 2014 15:18

Size of Africa300,000,000 smartphones coming soon. Carlos Slim of Telmex tells me the world is about to change. “Two billion more people will connect to the Internet when smartphones cost $50. The phone makers are promising me a $50 phone in 2014.” If Spreadtrum and Firefox deliver a $25 smartphone, as promised, that could accelerate takeover.

   ~310,000,000 Africans will be connected to the Internet in 2017, Arielle Sumits of Cisco predicts. The population of the U.S. is about 310,000,000, Africa over a billion. It’s inevitable that the U.S. will be dwarfed by the rest of the world. In Africa, there are already about 450,000,000 mobile phone users with substantial growth continuing. Most of them will get Internet-capable phones in the next few years. 

    There are fewer than 10M broadband landlines on the continent, about one line per hundred people.

Last Updated on Sunday, 02 March 2014 18:23
Vectoring Goes Live After 10 Years
Thursday, 20 February 2014 17:52

Belgacom world's first commercial deployment. Thousands of homes are connected at a solid, reliable 70 megabits down in Belgium, Ten years ago John Cioffi and George Ginis wrote the paper that introduced vectoring and it's been a long wait. Customers previously getting 30 meg down are now receiving 70 meg.  The DSLAMs are Alcatel-Lucent with Broadcom chips inside. The DSLAM are true nodes, not simple Board Level Vectoring. 48 lines true System Level Vectoring capable to vectorized 192 lines. The new home boxes (BBox 3)  are from Sagemcom and Technicolor, with Broadcom chips for vectoring. It wasn't easy, but the previous home box (BBox 2) with Ikanos chips has been made "vector-friendly."

   "Many but not all the lines up to 200 meters are testing at 100 megabits. We chose to be conservative and cap speeds at 70 meg. So far, the vectored lines are proving highly stable, G.inp and Seamless Rate Adaptation are proving helpful." Everyone's talking about 100 megabits from vectoring so I asked Patrick whether they will upgrade. "As we accumulate data from the rollout, we hope for a future move to 100 meg service."

Last Updated on Friday, 21 February 2014 14:03
OFCOM: Vectoring and Unbundling Can Work Together
Tuesday, 11 February 2014 07:29

Practicalities not determined. The British regulator in an updated paper declares "there are two ways in which vectoring and SLU may work together." http://bit.ly/MfNLwI They add that BT has already begun a trial and may very soon roll out vectoring, as I previously reported http://bit.ly/ZAGM3w. BT now seems to be wavering on vectoring. In technical committees, they are fighting competitors because they don't want to cede control.

   In the interim, British cable competitor Virgin was bought up by Fries and Malone at LGI, who strongly believe in tacit cooperation and higher prices rather than winning away customers. In half the country, BT has a monopoly because there is no cable. So they have little pressure to upgrade. In practice, European telcos tend to spend money to upgrade only when they fear competitors taking customers.

    Incumbents love the idea that vectoring imust be a monopoly and have been pushing that hard. Competitors strongly disagree. 

Last Updated on Thursday, 13 February 2014 11:33
FTTH Cost ~$1,000/home for Million Canadian Lines
Wednesday, 05 February 2014 09:35

Karen SheriffMerrill Lynch: "Right decision for the business." Eastern Canada's Bell Aliant intends to pass their millionth home with fiber by the end of 2014. The results in fiber areas are so much superior they are accelerating the build to 200,000 this year. This will cover 70% of the homes where they compete with cable. A few years ago, BA was being clobbered by Eastlink, an aggressive local cableco. They decided the only way to fight back was with fiber, even though majority owner Bell Canada wasn't doing the same.

   Glen Campbell of Merrill calculates they are spending $550-$600 per home, in line with Verizon's current costs for a similar build. Add the cost of connecting the homes and ~$1,000 is a reasonable estimate. Note that the next 30% of homes is predictably more expensive. Going from 70-90% wiould include maney homes that cost $2K to pass. The last 5-10% can cost $3,000-5,000 per, which is why even ambitious builds like Australia's don't intend to cover the last 5-10%.

   Kudos to Sheldon MacDonald and Karen Sheriff for demonstrating how efficiently fiber can be deployed. 

Last Updated on Wednesday, 05 February 2014 11:01
$200 for Fiber to the Basement in East Europe
Sunday, 06 April 2014 13:20

I didn't believe Rupert Wood's estimate but he has strong evidence. The going estimate for fiber to a typical neighborhood is $300-$500/unit, based on actual experience at DT, BT, AT&T and others. Analysys-Mason's Rupert Wood surprised me with the $200/unit estimate so I asked for details.

    This is an actual figure from several, not just one, Eastern European carriers, Rupert tells me. He shared enough details to convince me $200 corresponds to what these operators are seeing. Unfortunately, client confidentiality prohibits my publishing the specifics. 

~$1,000/home Fiber at CenturyQwest
Wednesday, 12 March 2014 18:15

The Enola Gay was built in Omaha 45,000 homes passed in Nebraska. Randall Stephenson of AT&T last year claimed "fiber costs have come all the way down" and Stu Ewing of Century just confirmed a figure of about $1,000 in modestly dense areas. "In Omaha, where we have done fiber-to-the-home to about 45,000 homes and it costs us about $600 per home passed to do that in Omaha. When you enable a house, a home and get the drop and the ONU and the set-top boxes, it’s $1,000 to $1,200 per home." http://bit.ly/1frXuH1 Obviously, fiber costs can be much higher. But Verizon, Bell Aliant and now Century are reaching the majority of homes at a relatively modest cost.

Last Updated on Wednesday, 12 March 2014 23:00
G.fast DSL has Momentum but G.hn Networking Wants Some Action
Monday, 24 February 2014 12:51

Upperside Conference ad

Production-ready G.hn/G.now silicon for fiber to the basement. The G.hn "home networking" folks are promising 200-400 megabit service 50 & 100 meters. They hope to steal some of the quickly mounting clamour for G.fast. (see pr at end) They are ready with high speed chips while real performance of G.fast chips remains speculative. This initiative, called G.now, shouldn't be surprising. The G.hn and powerline folk pioneered using frequencies to 100 MHz and 200 MHz, which is the heart of G.fast. Key G.fast engineers like Dudi Baum at Sckipio previously worked on powerline,

G.now's TDD architecture allows them to vary the split between upstream and down. Between both, speed "is 600 Mbps @ 50m, 500Mbps @ 100m, 350 Mbps @ 200m." They also promise bonding where the second line is available. That should almost double these speeds. They have Korea Telecom as a respected reference customer ready to begin. (Press releases below"

On the other hand, a senior engineer at a world-class telco writes me, "I do not foresee any service provider deploying a non-standard solution. For my point of view, it is not a threat at all for G.fast."

Last Updated on Thursday, 03 July 2014 17:16
Perlman's pCell Loaded with Hype But NY Times Calls 48 Megabits Over 100 Megabits of Spectrum Breakthru
Wednesday, 19 February 2014 08:29

Steve Perlman's pCell claims require a reality distortion field. Update: 2 professors confirm below.  Second Update. Met with Perlman and engineers. He has real engineers who did a slick trick using iPhones as LTE receivers. They may also have some neat tricks in MU MIMO and SON. Not close to their claims and most not new. Update 3: They don't give details about the software inside their server. It could be seriously interesting as improvements on how cells work together. Or it could be not.  More to come.  Original: Nick Wingfield in the Times believed a demo that could be reproduced on a cheap home WiFi router demonstrates a wireless breakthrough nyti.ms/1cZx3rS.  The Times showed 8 iPhones simultaneously playing HD video. Reporter Nick Wingfield wrote, "that would ordinarily bring a cellular network to its knees." Netflix streams at 5.5 mbps or less, so 8 streams is less than the 50 megabits even a modest cell site delivers, with 100-150 megabits becoming the standard rapidly. Within the space of a loft, speeds would be much higher. A 300 megabit router at Fry's this week costs $44.95 http://http://bit.ly/1ghadhU; Fry's is selling a gigabit router for $129.95 http://bit.ly/O9rWQM. Ericsson has demo'd 800 megabit LTE in a van driving around Stockholm. 48 megabits or even 400 megabits in a small space is not an advance.

    What Steve really has is a lot of hot air and what appears to be a prototype MU-MIMO system. If he has that production ready, that would be an important advance similar to work going on at all the major wireless vendors. That's similar to what Stanford researchers and many others have been working on for years. Literally in the last century, Stanford Professor AJ Paulraj described such systems. He predicted MIMO would one day produce a 1,000x and higher improvement in wireless speeds and that the theoretical limits could be 1,000,000x. Paulraj in 1993 invented MIMO, likely the heart of pCell.

Last Updated on Friday, 21 February 2014 20:50
France Telecom Wants Fiber to the Basement, Not All the Way Home
Saturday, 08 February 2014 16:49

ARCEP_FTTB Should France go gigabit or is 50-100 meg just fine. As part of the "incumbent takeover" spreading across Europe, France Telecom is fighting against government policy for full fiber home. Amidst pushback, ARCEP has called for a public consultation. (see below). Fiber home can deliver a reliable gigabit today, but Germany, Australia and England have decided the cheaper 50-100 megabits of DSL is all they will need for a decade and probably more. FT hopes to save a few hundred dollars per install, especially because they will draw power from the customers rather than the electric company. 

    Prettified with the name "fiber to the distribution point" (FTTdp), FT's plan uses puts a box in the basement and uses the copper to the apartment. They are already testing this in Poland with a box from Aethra, an Italian company using Lantiq VDSL chips. The Aethra box is single user, making vectored performance unlikely. There's no obvious way to connect the boxes for exchanging the information required for vectoring. As you can see from their release at bottom, they claim 200 megabits down although it looks like they are using VDSL2 chips normally considered 100 megabits. They did a demonstration at last year's BBWF of their box with 50 meters of direct wire. The speeds in the Youtube video were 227M down, 89M up http://bit.ly/1lrBz8g. Speeds in real world deployments will be lower. The boxes are a little clunky for now. I'm sure they (and their Lantiq partners) are working hard to reduce both size and cost. They need to get the cost well under $100 to be attractive for volume deployment.

   The Aethra box anticipates G.fast with "parasitic power." It draws power from the customer connection, eliminating the need for the carrier to connect mains power in the basement. The cost to the customer of that power is modest, perhaps $1-2/year. 

Last Updated on Monday, 10 February 2014 19:07
G.fast & FTTdp Get New Model from Lantiq
Wednesday, 29 January 2014 10:02
Actual line performance300 MHz signals behave differently from 30 MHz signals. Old models, such as the ETSI model for VDSL, do not accurately measure performance at higher frequencies. In the chart, there's a major difference from the theoretical curves predicted by older models and the actual performance measured by Deutsche Telekom. (The ragged plot of the actual measurements falls far below the smooth curves of the older models.)    

   Rainer Strobel of Lantiq,  Reinhard Stolle of Hochschule Augsburg and Wolfgang Utschick of Technische Universität München have built a new model that better fits the data. They write me "The industry is using a a cable model back from 1998, originally designed to describe cable properties for single line ADSL transmission. These models, adopted by ETSI and NIPP-NAI, served their purpose reasonably well for VDSL2  frequencies (<30MHz).    

   For the frequency range of interest for G.9701 (G.FAST), they are simply not accurate enough and don't reflect the physical coupling effects between pairs. For the purpose of developing G.FAST technology, we did require precise mathematical models of telecom cable binders up to 300MHz. These models have been developed and successfully verified against measurements of real world cables and did allow precise performance prediction and benchmarking of algorithms for G.FAST, that otherwise had never been possible with the 1998 versions."

    Alcatel's Stefaan van Hastel similarly observes, "the crosstalk behavior at these very high frequencies is very different than the behavior at the frequencies used by VDSL2. We're talking Vectoring 2.0. It's a significant change that needs to happen in vectoring" to work on the higher frequencies. http://bit.ly/1jokUEH He believes it will be years before the vectoring for G.fast is developed and it reaches maximum performance. 

    Everyone in the industry is waiting for actual results from production G.fast systems before coming to any conclusion on performance levels.

Last Updated on Sunday, 09 February 2014 20:18
More Articles...
<< Start < Prev 1 2 3 4 5 6 7 8 9 10 Next > End >>

Page 2 of 11