Template Tools
Deutsche Telekom, Telstra didn't know NSA had cracked them

Sckipio's little boxPossibly using spies and black bag jobs. Update: Further Snowden documents confirmed NSA used spies in Germany & Korea. The undoubtedly excellent engineers at Deutsche Telekom couldn't find how the NSA was tapping them. "Reporters for Der Spiegel, working in collaboration with The Intercept, contacted Deutsche Telekom and NetCologne several weeks ago in order to give them an opportunity to look into the alleged security breaches themselves. The security departments of both firms say they launched intensive investigations, but failed to find any suspicious equipment or data streams leaving the network." Those NSA guys have to be really good to fool the Germans.

    We all "knew" that surveillance was everywhere, but continuing revelations from Laura Poitras & Edward Snowden remain startling. The latest report suggests heavy use of undercover agents and physical intrusion. That's what spies do, after all. England's GCHQ is deeply involved, along with the Aussies and the Canadians. 

Let's make a deal: Verizon & AT&T looking for auction strategy

5G is comingPresident John Stankey and Verizon CFO Fran Shammo didn’t get together in an exclusive club and cut an auction deal.  They could go to jail. As a former Time Warner VP explained a while back, “We’ve become so good at signalling we don’t have to meet in airport motels anymore.” Stankey at a public conference said they would bid $10B in the 2015 auction but weren’t particularly interested in the 2014. Soon after, Shammo told Wall Street they were focusing on the 2014 auction and wouldn’t commit to even entering 2015.

   These are incredibly able executives likely to find a way to “bid rationally.” I don't see how I could prove anything without subpeona power but perhaps emerging star D.C. reporters Gautham Nagesh or Brian Fung might find a way. A remark by FCC Chair Tom Wheeler suggests he fears what I'm seeing; maybe someone at the FCC has perspective.

G.fast: Sckipio has the chips

Sckipio's little boxNot really a gigabit but 300-700 megabits is darn fast. Michael Weissman of Sckipio has delivered a highly integrated 4 port G.fast chip to a dozen device makers, making good on Dudi Baum's promise to have chips in 2014. He promises a CPE chip within six months and believes carriers will start deploying before the end of 2015. 

    "The interest is crazy," he tells me. Carriers that normally take a year or two to start thinking about things want to go to trials in a few months. "It's not just the European telcos," Weissman added. "We are seeing demand from Latin America, the U.S. and just about everywhere. Some want to go to advanced trials as soon as we can supply the gear." 

Dado's $22M, Alcatel's $15M buy Ikanos time to finish chips

Provides time for node scale vectoring, G.fast. I'm very glad I don't have to write an obituary for Ikanos, home of a large team of respected engineers. $22M from Dado Banatao's Tallwood will keep Ikanos afloat, but it's the $15M Alcatel is putting in that points to a bright future.

    Alcatel remains the largest DSL vendor and has long bought from Ikanos, but many of their DSL chips have been coming from Broadcom. With 40M lines of vectored DSL on order in Europe and Australia, Alcatel wanted to make sure they had several suppliers. Ikanos has been developing a vectoring chip aimed at AT&T and other telcos that want to vector 192 ports in large cabinets. 

    Last year, I killed a story about Ikanos although I was confident it was accurate. Unfortunately, the chips they needed were taking longer than hoped. I listened to the CEO on the conference call and ran the numbers. I was convinced they were soon about to run out of money. There was a registration for a new financing but it wasn't clear whether anyone would put the money up. 

    I don't overestimate my influence but realized that if I published the article it might be the final straw for the company. I decided I didn't want to be responsible for putting 100 hardworking people out of work in a company that had earned my respect. This is not standard journalism practice but as owner I make my own choices.

On privacy, "conflicting feelings" of the Internet governors

Apple, Google & Cisco "support" privacy but Internet leaders are conflicted. "FBI Director James B. Comey sharply criticized Apple and Google on Thursday for developing forms of smartphone encryption," the Washington Post reports http://wapo.st/1qHVVtQ. I believe the FBI is fighting against a rough consensus among the IETF and Internet technical community that privacy should be protected by technical means. That includes encryption that neither the government nor the cell phone designer can crack.

    Bob Hinden, Chair of the Internet Society and a long time leader at IETF etc. sent Dave Farber's IP list an interesting note. I'm picking it up because I've heard similar from others, not to single out Hinden. 

"Dave, I admit to have conflicting feelings on this.  Part of me is happy that no one has access to the keys, but at the same I time I have some sympathy for legitimate law enforcement.  There are bad guys out there.  If I had to choose, I would pick civil liberties over law enforcement access.  With what we have learned about what the NSA is doing, I think it's time for the pendulum to swing back more towards the middle.  NSA (with the apparent support of Congress and the Administration) clearly has abused their capability to get the keys.


    I'm glad Bob chose to support civil liberties. Many others make different choices. Cisco, for example, is not merely very close to U.S. security agencies but played a crucial role building the Great Firewall of China. I was glad to hear Cisco support encryption recently but remain skeptical. It's almost a joke the U.S. reps on Internet security bodies including the ITU come from Cisco.

China: 60M Fiber home connections 135M homes passed

Possibly as many as the entire rest of world and soon pulling ahead. Government owned and controlled Chinese telcos are building more fiber home than the entire rest of the world the last few years. The 60M homes connected http://bit.ly/Zh6AX7 are not far from the total fiber lines in the entire rest of the world. The OECD includes most of the developed world and counted 56M homes with fiber at the beginning of 2014. There's a fair amount of fiber in places like the UAE so I'm guessing "ROW" remains slightly ahead of China. That won't be true for long.

     Dell'oro, one of the top research houses, reports "PON revenue in Q2 reached a record level, with strong growth both in China and the rest of the world." By revenue, 60% of the sales were in China, up 25% over last year. Since Chinese prices are lower, an even larger % of the ports went to the Middle Kingdom. GPON in China is taking over, with EPON sales flat. Strength in China made Huawei and ZTE the world leading vendors. 

The marginal cost of bandwidth: Under $1/month in NY, 80X higher in Lagos

Perhaps $0.18/gigabyte in Sao Paulo, $2 in West Africa. Bandwidth is not free but it's cheap and getting cheaper. The latest figures from Telegeography, the industry standard source, are that high volume transit costs less than $2/megabit in New York & London, perhaps $6 in Hong Kong $18 in Sao Paulo and $170 in Lagos. "IP transit prices in New York and London are among the cheapest in the world, with median prices of $1.64 and $1.36 per Mbps per month, and the lowest prices at less than half this. ...  At USD18 per Mbps per month, the median 10 GigE price in Sao Paulo is eleven times higher than in New York and 13 times above London. At USD6 per Mbps per month, prices in Hong Kong are more than three times those in New York and four times those in London." I separately checked the price in Lagos. It is $170/megabit/month in substantial quantities. That's literally 100 times more than London. Some of the difference is the cost of the undersea cable but most relates to cartel market power.

   Argentina, Brazil and Chile October 2 submitted a proposal for the ITU Plenipot to do something about the high costs. (Below) The U.S. will probably oppose this, with AT&T on the U.S. delegation. U.S. State generally supports U.S. businesses against the needs of developing nations, discrediting all our efforts in Internet governance. 

    Backhaul costs are quoted in $/megabit.month. For policy purposes, that needs to be translated into $/customer/month. That's affected not just by total demand but by the pattern. You have to design a network for the peak, not the average, usage. It's probably accurate to divide the  cost per megabit by 5-10 to a figure for the cost per customer for transit, etc. That's 16 to 30 cents in New York.  

    The carrier then has to carry the bits from the peering point to the local DSLAM or CMTS. Conveniently, the on network cost is similar to the the cost of transit. Both involve 3-7 hops. Each hop requires a router; connecting a hop requires fiber and the gear on each end of the fiber. So you get a reasonably accurate cost per customer by doubling the transit cost. That yields 30 cents to 60 cents per month per customer to a carrier like Verizon or Time Warner. I prefer to use a figure of "less than $1/month" to cover all the uncertainties in the calculations. $1/month/customer corresponds to the marginal cost figures per capita used for rough estimates at some large carriers. That would be more reassuring if I were not virtually the only source publicly estimating the figure; I could be picking up an echo of my own calculations, as I discovered I had when quoting BellSouth on similar years ago. 

It is important to realize the costs faced by many rural and smaller carriers are much higher. 

NY Times doesn't get that MIMO increases spectrum capacity 300% and more

Lowell McAdam knows spectrum Five years of WiFi prove multiple antennas dramatically improve throughput. A key reason D.C. policy is awful is that many in D.C. don't realize the extraordinary increase already happening in spectrum capacity. The engineers are instead debating whether the MIMO increase will be 10x or 100x, with the higher estimates getting strong support. http://bit.ly/MSDCREG

    Four years ago, most U.S. cell phones ran at 1-3 megabits. The latest measurements have AT&T at 14 megabits, Verizon at 16 and T-Mobile at 19! Even a reporter can see the networks are more than keeping up and the lobbyist spin is highly implausible. But Wyatt reports "mobile broadband depends on the public airwaves known as spectrum, which is a finite commodity with limited capacity," based on a comment from Meredith Baker, D.C. lawyer/lobbyist.  The actual problem for the telcos is finding enough customers, not finding spectrum.

    That's just the start; Spain, Singapore, Korea & the Emirates are deploying 300 megabit shared LTE, which is twice as fast as the U.S. offerings. Nokia is demonstrating 10x that speed, 3.8 Gigabit LTE. 3.8 Gigabit? Just A Demo, But Definitely Feasible http://bit.ly/38gigabits If you want 10 gigabits, Ted Rappaport just wrote a book explaining how that will be delivered with millimeter waves, perhaps 28 GHz. http://bit.ly/Millimeterbook

    Instead of believing Baker, a lobbyist, Wyatt should read what Verizon CEO Lowell McAdam & AT&T President Ralph de la Vega are telling investors. AT&T & Verizon: We Have Enough Capacity So Spectrum Isn't A Big Issue http://bit.ly/Bellsspectrum Lowell, an experienced engineer, notes "We have a lot of AWS spectrum, We have great capacity in place. ... the network has the capacity." In fact, Lowell has been cutting capex while speeding up the network. "We expect to see capex as a % of revenue to fall."

To Tap a DSL line, use Tracespan

Tracespan interceptWikileaks isn't all secret. Tracespan makes some of the best DSL test gear, so I shouldn't have been surprised to see them on a list of tappers at Wikileaks. Wikileaks' SpyFiles is making news with details of high-powered hacking tools shipped from the West to governments which abuse dissenters such as Egypt, Qatar & UAE. They identified $65M in FinFisher "weaponed malware;" there's plenty of money being made here. Alcatel, Siemens and the other western communications vendors were heavily represented, some leaked.

   I also found in the Wikileaks files the Israeli company Tracespan. The Wikileaks data on Tracespan was interesting but not secret. I found some of the files at http://www.tracespan.com/ipVDSLPhantom.aspx under "lawful intercept." I'm sure Tracespan cooperates with Israeli intelligence, which like the U.S. CIA or French sécurité extérieure is a "by any means necessary" outfit. (Think Iran-contra.) 

   The U.S./Australia Huawei ban may inspired more by the difficulty for U.S. spies than the danger from Chinese ones. 

Doug Sicker moves from Colorado to Carnegie-Mellon

Doug SickerBoulder Colorado remains a top telecom hub. About a dozen engineers/academics play an outsized role in D.C. policy, having earned friendship and respect from the top policymakers. Sicker was the Chief Technologist of both the FCC & NTIA  and is always heard when he goes to D.C. (Doug's picture is from long ago.)

   Carnegie is one of the world's best computer science departments but I was surprised be moved to Pittsburgh. He loves Colorado.  I asked Doug why he moved and his reply is below.  

   The University of Colorado at Boulder is among the very strongest centers for telecom work, with Phil Weiser (close to Obama) and Dale Hatfield (D.C.'s longtime favorite engineer.) David Reed has recently come over from CableLabs to run the Interdisciplinary Telecom Program.

For the record: Dave to FCC on what to do when competition doesn't work

Will Wheeler do anything to reverse the high U.S. prices? That would be the logical followup to his speech asserting  “meaningful competition for high-speed wired broadband is lacking and that Americans need more competitive choices for faster and better Internet connections.” http://bit.ly/FCCBlog I'm skeptical and commented on the FCC blog:

Wheeler is going in the right direction but both of you are ducking a crucial question: What's the right policy when competition is weak and there's no practical way to "incent competition." When I go to Wall Street, nearly no one seems willing to invest the $billions if would take to broadly take on the incumbents, no matter what "incentives" you find. Competition is great but when it's not working we need to be realistic. 

   Over the last 7 years, U.S. prices for high-speed broadband have risen to typically 30%-70% higher than Germany, France & England. 

   Time to do more than talk and pray. 


Live G.fast at Telekom Austria

Telekom Austria A-1"A customer" on FTTB/G.fast. Peter Schiefer of TA confirms, "We are going to show G. Fast with a live customer next week on Wednesday." If it is just "a customer," Alcatel may have installed a lab test style rig with FPGA's. It could be an early Broadcom chip Alcatel is testing. Broadcom as usual is keeping mum and Alcatel is offering no details before BBWF on what they have in G.fast.

    Austria is talking a billion euro subsidy for faster broadband and just allocated the first 300M. That gives TA powerful incentive to showcase advanced capabilities. Management needs to look progressive. Carlos Slim already has three seats on the board and appears to be moving ahead with a takeover. Slim's Telmex knows how to run a low cost operation, crucial as Austria expands into Eastern Europe.

     Telekom Austria operating in Serbia still is a surprise 100 years after Franz Ferdinand. 

2014-2015: 230M iPhone 6

Despite worrisome low demand in France, China. Apple is looking to build 116M 6's by about the end of January and as many again in the following 12 months. That's a remarkably ambitious plan for an expensive phone. Some very good phones are appearing at half the price. The figure comes from NPD/Displaysearch and is based on projections from the companies that supply screens to Apple.  

    Apple stands alone rising above massive upheaval in mobile. The sales rankings in cellphones are shifting rapidly.  Samsung just gave a warning and previously hot Coolpad laid off 10%.Xiaomi has expanded from China to Indonesia & India and selling very well.  In India, I'm watching Micromax, Karbonn & Spice with climbing sales.  ZTE was hot, lost a lot of money, and then fell back. Blackberry was on top of the heap a few years ago, but only a huge cash hoard has kept them from bankruptcy. That Apple will continue to do well seems a safe projection, but nothing is this field is certain. That they will pull even further ahead, as these figures imply, is amazing but corresponds to much of the early data. 

   "It's still the best smartphone you can buy," the WSJ reviewer concludes. Apple fanboys please don't hate me, but I have to note the iPhone 6 may be already out of date. LTE-A is going to two & now three carrier, 250-450 megabit speeds in Spain, Germany, Singapore, UK (EE) and probably Sprint in the U.S.

Canada DSL Takes 2% from cable: U.S. 2% the other way

With most of Bell & Telus upgraded, Canadian cable broadband homes almost go negative. Glen Campbell of Merrill Lynch pointed me to CRTC data http://bit.ly/1voSz5q that showed the Canadian telcos beating the cablecos, who lost 1.3% of the market in 2013.  In the U.S., the results were the opposite, with cable picking up a point or two of market share. Telcos are beating cable in Canada, Britain and France. Cable is beating telcos in Germany and the U.S..

    The largest single factor is the % of homes upgraded by the telco. A strong majority have been upgraded in Britain and Canada, but fewer in the U.S. In Germany, cable is competing by offering twice the speed at the same price, easy because DT is only now getting serious about upgrading. Mike Fries and Liberty Global/Unitymedia KabelBW prefer to raise prices than to increase market share, so we'll see how that plays out.

    The U.S. telco figures are skewed by the AT&T/Verizon decision to kill all landlines to the majority of their territory, going wireless only to 20-30% of homes. They are treating those homes like the Romans treated the Sabine women so of course many are fleeing to cable. Where they have FiOS and U-Verse, they are doing fine, but in the other territories AT&T is doing so poorly they lost a net 50,000 lines in Q2.  

Aintnecessarilyso: Faster broadband yields big income increases.



Author Erik Bohlin says his work on an income effect "was very preliminary, needed to be replicated, and are not ready to use to make a policy discussion." A quick read of the work showed it was extremely limited. So I was surprised to see an OECD paper http://bit.ly/1sBoqRB by a distinguished scholar presents as fact, "increasing broadband speeds from 4 Mbps to 8 Mbps produced an increase in household income of USD 122 per household per month, and increasing from 8 to 24 Mbps produced a significant further increase beyond that figure (Bohlin et al, 2013)." That's definitely unproven. 

   I find it highly unlikely that watching a second stream of Netflix or more rapidly making a Tinder hookup yields such a large economic payoff. Common sense would require being very skeptical about such a claim and looking closely at the data before using it. Serious scholarship has questioned whether there really is an economic payoff from broadband; in several cases, if there was a net economic payoff it was too small to measure.

     I blame myself for the OECD sloppiness because I knew the Ericsson/Bohlin work had severe limits and never got around to writing an article. The original work was funded by Ericsson, maker of broadband equipment, another red flag. I am carefully not claiming Bohlin distorted his findings in return for Ericsson's money. In a follow-up with me, Bohlin and co-author Ibrahim Kholilul Rohman have been sensible about the limits of their work.

BT Faking it on G.fast

200-600 meg pretty fast so why hype? British Telecom, despite the BBC and other news reports, did not attain a gigabit with G.fast. They merely showed that 104 MHz over a protected standard copper line could reach speeds of 786 megabits down 19 meters, something obvious for years from the basic engineering. The substance of G.fast is significant so the hype is unneeded.

BT's results were not for G.fast. G.fast is a 300 page specification that BT/Huawei didn't even pretend to support. Many features, such as reverse power, are important and were not implemented. Vectoring in G.fast remains totally unproven and will be crucial to any deployment. Several features of G.fast are likely to reduce the speed. The effect may be relatively modest but we won't know until they are tested.

10 gigabits + using higher frequencies

RTed and Robert's new bookJapan, Korea plan 28 GHz 5G for 2018-2020. If 3 gigabits isn't enough, in a few years you'll be able to go to "millimeter wave," 15 GHz, 28 GHz, 60 GHz WiGIG or similar. There are literally gigahertz of spectrum available, enough for 10 gigabit and higher speed service. At the Brooklyn 5G Wireless Summit, I spoke with AT&T President John Stankey and NTT DOCOMO Seizo Onoe.

    Onoe is ready go as soon as the equipment is ready, with Alcatel-Lucent, Ericsson, Fujitsu, NEC, Nokia and Samsung supporting his trials per the release below.  Stankey is more cautious, still studying, but he wouldn't have come to Brooklyn if AT&T wasn't serious. Lots of problems still to solve, including short reach and difficult reflections, but there are hundreds of engineers at major companies already at work. 

   The "Prince of 28 GHz," Ted Rappaport of NYU Wireless, led that conference and has a new book out, Millimeter Wave Wireless Communications. It's certain to become the primary reference because of the prominence of the authors. His co-author Robert Heath has built an MU-MIMO testbed at the University of Texas that's a crucial proof of concept for MU-MIMO. It's professionally priced ($100-125) but through Thursday 9/25 you can get a 40% discount with the code MMWAVE40 at http://bit.ly/1Dmi5fk

3.8 Gigabit LTE. 3.8 Gigabit? Just a demo, but definitely feasible

Nokia demoMasayoshi Son at Sprint or Korea Telecom in Rwanda could deliver a gigabit in a few years. Nokia used 200 MHz, about 5 times the spectrum advanced carriers are using today, and achieved about 15 times the performance of today's better networks. Nokia has just demonstrated 3.8 gigabits, as you can see in this video. http://bit.ly/Nokia38

    A team at Rice & Cornell Universities have designed a chip they believe will support a similar 3.8 gigabits in only 100 MHz, the amount of spectrum available to Sprint as well as most of Latin America and Africa.  (Note below) Henry Samueli of Broadcom in a Marconi webinar predicted chips like that. Since 2011 and 3GPP release 10, every informed wireless engineer has known speeds will pass a gigabit (shared) on many commercial networks.

   That kind of performance is just what we'd expect as MU MIMO and massive MIMO come out of the labs. Some excellent engineers on Oct 2 in D.C. will be discussing MIMO 2025:  A 10x or 100x Capacity Multiplier? at the Marconi Symposium.  Stanford Professors Paulraj (inventor of MIMO) & Cioffi, Robert Lucky and Thomas Marzetta of Bell Labs, as well as Helmut Bölcskei of ETH Zurich are on the panel. http://bit.ly/MarconiDC  If you're in D.C. do attend; elsewhere, the webcast will be at http://marconisociety.org/blog/index.php/symposium-live-stream

   Re-farming alone should yield about 100 MHz in most countries, even without adding more monopoly spectrum. Good engineers debate whether the performance increase will be 25X, 50X or 100X today's average network, as long as the regulators are neither stupid nor corrupt.  

Soon come: 145 MHz spectrum, 3 gigabit speeds in Rwanda

Rwandan ORN workersKorea Telecom/Rwanda Government ORN spectrum could support ~2 gigabits across the country. While the initial rollout is using only 20 MHz, the Nokia equipment being installed can expand quickly to 60 MHz/400 megabits. There's a clear roadmap at least to 100 MHz with higher MIMO, enough to go well above a gigabit.  Nokia demonstrated 3.8 gigabits at the recent Mobile Asia event, using 200 MHz. http://bit.ly/38gigabits

   Joint venture ORN has 145 MHz of prime low band spectrum between 700 MHz and 900 MHz. This is one of the world's first wide allocations of spectrum, the most efficient way to increase capacity in the LTE and 5G era. Large spectrum blocks are the right move for capacity in any country with few wires, including India, Indonesia, and almost all of sub-Saharan Africa. 

AT&T & Verizon: We have enough capacity so spectrum isn't a big issue

Ralph de la Vega & Lowell McAdam are pros. More spectrum makes adding capacity cheaper but any wireless engineer can tell you the "spectrum crisis" is political bunkum. This week at Goldman Sachs, CEO McAdam of Verizon and President de la Vega of AT&T made clear they can move forward whether or not they get more spectrum. That confirms what the previous Verizon CEO Ivan Seidenberg and AT&T's John Stankey had said in 2009. Both were comfident they'd do just fine. Five years later, their average customer speed on LTE is about 15 megabits, ten times as high as 2009. 

    Ralph, one of the most respected executives in the business, told investors "We feel really good about where we are. ... we don't have this burning desire for the need of coverage or for other reasons to go aggressively after Wi-Fi. Other operators that may have less coverage or have other needs, may pursue it more aggressively." Both Verizon and AT&T have boxed in the unused capacity of their customers DSL & fiber connections. Turning it on would add ?30%-50% more capacity in territory.

    Lowell, a trained engineer, "We have a lot of AWS spectrum, We have great capacity in place. We’re densifying the network with either small cells or LTE unlicensed. So the network has the capacity.  ... We have the assets in place. I don’t think we need a heckuva lot more. ... It goes a long way beyond just getting the spectrum. ... Always looking for efficiency. The small cell technology being deployed gives us a lot more capacity. ... WiFi is a critical part of the ecosystem managing the network. Using unlicensed spectrum is going to be important for us as we go forward. We intend to deploy LTE-U with the small cell technology, integrate it within the wider macro network, There are many dynamics that are involved in being more efficient. ,,, Even working with content providers to make sure you have the right formats to put less load on the networks."

    "We expect to see capex as a % of revenue to fall. I hope it continues at least at the absolute level," confirms that Verizon is not squeezed.

Stankey of AT&T: 5x5 spectrum blocks just aren't right

Rwandan coffee picker from Borlaug InstituteYet another conflict between efficiency and competition. At the coming U.S. AWS-3 auction, several of the lots are 5 MHz down and 5 MHz up, in order to let smaller companies bid. Small spectrum allocations are highly inefficient, cost more to build, waste some spectrum to avoid interference and possibly require special handsets to "aggregate" different bands.  The LTE-A that's deploying today is most efficient with 50-100 MHz of contiguous spectrum, not 5 MHz or 10 MHz. Older versions of LTE were designed around 10x10 bands. Competition is great when it works but the "cost of competition" is often very high. It's much cheaper to build one network instead of four or even two. lines.

<< Start < Prev 1 2 3 Next > End >>

Page 1 of 3