& cplSiteName &

Sckipio Pitches G.fast Boost as Telco Weapon Against Cable

Iain Morris
5/16/2016
50%
50%

G.fast chipmaker Sckipio hopes a new bandwidth technology will improve the business case for G.fast by giving telcos an edge over their cable rivals on upstream performance.

The Israeli startup has begun demonstrations with broadband equipment vendor Adtran Inc. (Nasdaq: ADTN) of a technology it calls dynamic bandwidth allocation (DBA), which is already being used in passive optical networks to share bandwidth between users.

In the G.fast case, however, DBA has been designed to boost the upstream bandwidth available to customers on copper-based telco networks.

Upstream speeds are still relatively low on this infrastructure, as well as the DOCSIS networks used by cable companies. In the UK, for instance, the premium residential service from telco incumbent BT Group plc (NYSE: BT; London: BTA) promises 76 Mbit/s on the downstream but does not go above 19 Mbit/s on the upstream. The best that cable competitor Virgin Media Inc. (Nasdaq: VMED) can offer is 200 Mbit/s on the downstream and 12 Mbit/s on the upstream.

BT has been an active proponent of G.fast, which extends the frequency range to boost connection speeds over last-mile copper links. Using G.fast, it hopes to increase downstream connections to as much as 500 Mbit/s. But without DBA, upstream speeds may continue to lag. (See BT to Cover 2M Homes With FTTP in $8.7B Plan.)

At the moment, the chief market requirement is for a good downstream connection that allows customers to watch movies and access more advanced Internet services. But this could soon change with the growing popularity of cloud computing, interactive gaming and other user-generated content services.

This is the opportunity that Sckipio Technologies is targeting. Rather than giving networks any kind of always-on "symmetric" capability, the DBA technology it has developed works by dedicating more capacity to either the upstream or the downstream as circumstances dictate. If, say, a user begins uploading a number of video files to the cloud, the upstream capability would automatically be increased to cope with that demand.

The trade-off in this example appears to be that downstream capability would diminish as upstream bandwidth is increased. In other words, watching a movie in high definition while the video files are being uploaded might not be a straightforward option.

That said, Sckipio is making some bold claims for the DBA technology, saying it could support connection speeds of up to 750 Mbit/s in each direction on today's chipsets.

It seems unlikely to be available anytime soon, though, or even exactly in its current form. Sckipio expects many elements of DBA to find their way into an International Telecommunication Union (ITU) standard called dynamic timeslot allocation (DTA). That technology will be the one intended for commercial deployment, rather than just demonstration purposes.

"[DTA] is being vigorously discussed now and may or may not make standardization this year," says Michael Weissman, Sckipio's vice president of marketing. "I'll be honest -- there are complexities in it and we want to take into consideration everyone's thought processes and create the perfect solution."

Nevertheless, Weissman argues the work that Sckipio has already done on DBA gives it a DTA head start of between six months and a year on Broadcom Corp. (Nasdaq: BRCM), which he evidently regards as Sckipio's biggest G.fast rival.


For more fixed broadband market coverage and insights, check out our dedicated Broadband content channel here on Light Reading.


What's clear is that, as a technology concept, DBA has already piqued the interest of telcos plotting G.fast deployments.

In a video interview about G.fast with Light Reading in April, Eddy Barker, the vice president of technical design and architecture for AT&T Inc. (NYSE: T), flagged interest in using DBA technology to provide much higher-speed upstream services to customers. (See AT&T's Big Plans for G.fast.)

Moreover, an endorsement of DBA from Canada's Telus Corp. (NYSE: TU; Toronto: T) is included in a Sckipio statement about demonstrations of the technology at the G.fast Summit in Paris this week. "DBA is a G.fast game-changer," said Tim Fell, the vice president of video and broadband services at Telus, in that statement. "In the race to deliver ultra-fast broadband, the ability to offer affordable symmetrical services will give telcos the flexibility required to meet our customers' evolving high-speed Internet needs."

None of this necessarily means that AT&T or Telus are making huge commitments to G.fast or Sckipio, but it could bolster the appeal of G.fast to operators weighing their broadband technology options.

Following recent improvements, G.fast already appears to have convinced a few big telcos they can sweat their legacy copper assets instead of rolling out new fiber. But critics have argued that G.fast is simply prolonging the life of a dying technology, and delaying an inevitable fiber upgrade in the years ahead. (See NBN Looks to G.Fast to Reduce Fiber 'Hassle'.)

Right now, the big question is whether DTA really will give telcos an upstream advantage over cable operators. Of this, Weissman sounds fairly confident. "DOCSIS 3.1 is the next three-year [cable] roadmap and so unless they can do something really quickly, this is going to catch them flatfooted," he says.

Yet even if DBA is not on the cable roadmap, symmetrical capability is. Indeed, Finland's Nokia Corp. (NYSE: NOK) today announced demonstrations of a symmetrical 10 Gbit/s service over a coax connection using Full Duplex technology. CableLabs and Kumu Networks have also been exploring Full Duplex, which shares the same frequency bands between upstream and downstream signals to support symmetrical services. (See Nokia Demos 10-Gig Over HFC.)

Given the real-world challenges surrounding Full Duplex, however, it will probably not see commercial deployment until at least 2020. If the G.fast community can bring DTA to market in the next couple of years, operators of copper-based networks may have something to shout about.

— Iain Morris, Circle me on Google+ Follow me on TwitterVisit my LinkedIn profile, News Editor, Light Reading

(17)  | 
Comment  | 
Print  | 
Newest First  |  Oldest First  |  Threaded View        ADD A COMMENT
Page 1 / 2   >   >>
MrFinance
50%
50%
MrFinance,
User Rank: Light Beer
5/19/2016 | 6:16:07 PM
Re: Time to market and playing catchup
I see what you mean on video.... Maybe... increasingly Netflix etc let set my own stream quality products preference and if my broadband supports it I'll go high end even if I'm not holding my phone at a 50inch equivalent distance.. Maybe that's just me though. Regarding on-live, I'd see their failure as more business model than anything else. They should have sold to Microsoft early on, and anyway the game streaming business model was crappy when so few consumers had broadband speeds well in excess of 20-30mb/s to support it. I don't think that will hold true by 2025 and it may be a far superior option for Sony to kill the physical console generations off at the ps4k and release PlayStation 5 in the cloud and get the immediate audience of 95% of the ps4 / ps4k owners who will have adequate bandwidth by then. Anyhow I think we are of the same mind that massive speed increases to GB and certainly beyond will not be economic for consumers to pay for, let alone need. Thanks for the interesting thoughts.
brooks7
50%
50%
brooks7,
User Rank: Light Sabre
5/19/2016 | 10:36:09 AM
Re: Time to market and playing catchup
Except that GPON doesn't have a need for DBA downstream as each customer can consume all the bandwidth needed as there are no timeslots.  So, it does provide exactly comparable functionality.

As to whether symmetric is super important....maybe.  The ability for a home to upload videos (and remember we are talking about home users without any business applications) is interesting but nobody is going to be able to charge for it and I am confused on how the marketing for that wil go...."Hey upload your sex tape to Youtube at 10x the speed!"

I see g.fast as a nice little niche product that will go into a smaller market than VDSL2 which in itself was a nice tiny niche.  There are so many places where the technology doesn't really work economically due to housing density.  What do you do in rural where it will be 1 home passed per dslam?  It seems to be a good technology for apartments with a dslam per floor, but in standard density housing it is not clear that this is a good fit.

seven

 
spawnbsd
50%
50%
spawnbsd,
User Rank: Lightning
5/19/2016 | 6:33:56 AM
Re: Time to market and playing catchup
@brooks7

DBA on GPON has nothing to do with DBA on G.fast, the name is a poor choice which is why many G.fast vendors refer to it as DTA now. DBA on GPON allows for dynamic allocation of timeslots in the upstream direction (only) for each subscriber on the PON, based on actual subscriber traffic. DBA/DTA on G.fast allows for dynamic allocaton of timeslots in the upstream AND downstream direction based on actual subscriber traffic; unlike the fixed upstream/downstream allocation in VDSL2 due to FDD usage.

Independent DBA/DTA for coax use cases, allows the upstream and downstream timeslot allocation to change uniquely for each subscriber line. Collective DBA/DTA for twisted pair vectored use cases, allows the upstream and downstream timeslot allocation to change dynamically for all subscriber lines; at the same time (this is required, otherwise we get killed by NEXT). Obviously independent DBA would be preferred, but collective DBA would still give 'burstable' symmetric capabilities, especially on lower density DPU's where it's statistically improbable two subscribers would be uploading and downloading at the same time (and exhausting all timeslots allocated).

DBA/DTA on G.fast is important as it gives burstable symmetric capabilities on copper, something the cable co's with D3.1 will not able to respond to. Yes full duplex is coming to cable, but despite CableLabs thinking it's a small change and will be possible soon, it requires significant chipset redesigns and cable plant changes. By the time CableLabs has full duplex on D3.1, we'll have 3rd or 4th gen G.fast chipsets, which will also support full duplex on twisted pair.
brooks7
50%
50%
brooks7,
User Rank: Light Sabre
5/18/2016 | 1:13:51 PM
Re: Time to market and playing catchup

"After five years of uninterrupted service, the OnLive Game Service comes to an end."


 

Streaming video services have not worked.  There are lots of reasons for that which has to do with the business models in the gaming community.  And the technology that I have mentioned has been around since Ultima Online and Everquest and is used today in Call of Duty, Counterstrike and World of Warcrat.  

No...video on a 50 inch monitor takes more bandwidth than on a 10 inch monitor for the same quality from the viewers perspective  As devices become more personal and shared devices decline, I think we will see a slow capping of entertainment content bandwdith.  Note, we are still talking about something 10 years from now and can be handled by today's technology as deployed today in the fiber and coax realm.  GPON = 32 customers per 2.5Gb/s.

I am not saying that bandwidth usage won't continue to grow, but there is this limit of cost per home and bandwidth per home in the access plant.  The plant is a 1 to 1 relationship in expenditure.  So, how much are service providers willing to pay if they can't charge any more money for these exploding bandwidths.  Right now a GPON ONT is about $100.  That is the scale we are talking about for Gigabit deployments.  Let's figure out when a 10Gb/s ONT can be about $100 and when end devices in the home will have 10GbE ports on them.

seven

 
MrFinance
50%
50%
MrFinance,
User Rank: Light Beer
5/18/2016 | 12:16:45 PM
Re: Time to market and playing catchup
Not quite...I'm referring to cloud game streaming (on-live, Sony Gaikai), which unless vr takes off will scale to much better visuals/price than console/pc gaming provided broadband can take c. 30mb/s at 4k equivalent (the compression is not quite as efficient as video as it needs to be done in real time for acceptable latency for screen based gaming - not possible currently to do compression at acceptable latency for vr based gaming). Not sure I follow your point on personal - do you mean smartphone and tablets tend to be lower resolution than TVs? If so, then maybe, depends a bit if flexible screens get perfected in which case most devices will be 4k by 2025
brooks7
50%
50%
brooks7,
User Rank: Light Sabre
5/18/2016 | 9:19:45 AM
Re: Time to market and playing catchup
 

The mistake you have in your math is that gaming is more like 56K.  The tech for online gaming is still based on that which was used in dial-up.

And as screens get more personal the bandwidth needs go down.

seven

 
MrFinance
50%
50%
MrFinance,
User Rank: Light Beer
5/17/2016 | 5:43:59 PM
Re: Time to market and playing catchup
I take your point in the short term (actually it's the same point i was making - speeds today are mainly marketing not use driven). That said, i think that 100mb/s is a realistic future sustained usage for a few hours peak if not 24/7. 2025's peak hour 5 person western house= 3-5x 4k security camera and/or userbroadcasting upstreams = 60mb/s + 3x (5users household,, 2 of which sharing a screen) 4k content downstream (gaming or video) = 45mb, + a few random background software downloads from laptops/phones/smartcars/ smarthealth/ smart whatever for another 10-20mb (as mostly it will be set to download outside peak). You can maybe bump that up a bit if you are a big believer in VR, in which case double the download streams to 8k equivalent content and eliminate the screen sharing = 150mb/s download and c 2-250mb/s total. Cut that a bit if you think h265 compressors will get better or recognise that vr is so proprietary still that we'll likely end up using something better than h265. Whatever you believe, for sure it's not a GB/s sustained. The only thing that could really need fibre rather than copper is uncompressed VR cloud gaming where the streams need to be 6-10gb uncompressed to hit the latency needs. Personal view though is that (even setting aside the fibre physical capex cost), the chances router/switch/optics costs for cloud vr gaming scale down faster than graphics card costs to generate the gaming locally is pretty much zilch. Hence I'm not a big fiber believer. At least in europe, copper can carry the sustained usage and you can hybridize the home gateway with mobile to cover off the marketing claim need vs cable and fibre. I'm not a us expert, but even with your worse copper topology and smaller mobile spectrum allocations, would worry fibre business cases will struggle.
brooks7
50%
50%
brooks7,
User Rank: Light Sabre
5/17/2016 | 5:09:10 PM
Re: Time to market and playing catchup
Not 100Mb/s per customer.

100Mb/s per customer constant usage.  Meaning that the 100Mb/s is always on.  Today that is less than 1Mb/s on average.  Think about consuming 1M Megabytes of information each and every day. Not 1 Megabyte, 1,000,000 Megabytes.  That is about 100Mb/s of constant traffic.  Call it 3000 hours of HD video downloaded each day.

seven

 

 
MrFinance
50%
50%
MrFinance,
User Rank: Light Beer
5/17/2016 | 5:00:52 PM
Re: Time to market and playing catchup
I hear you, telco is an odd mix between a tech and a capital asset business... My counter would be you can depreciate your old assets on lower price tiers and you should be building that price decline and user canibilisation into your initial return on capital assumptions lest someone (cable, fibre over builders or 5G startups) spots the invitation to eat your lunch. Of course lots of telco is not fully competitive and continues to earn returns well in excess of wacc, so I'm not arguing that you do get the behaviour you suggest and that it's appropriate behaviour from the point of view of the investors if not the customers or government. PS they'd better upgrade from 100mb to GB fibre if they want to beat d3.1 offers. I'm fully with you that it's largely pointless from consumer use case for the foreseeable future, but 1gb is such a pretty little marketing term no, and maybe I've got the math wrong but strikes me the upgrade costs are a lot less than the marketing hit.
brooks7
50%
50%
brooks7,
User Rank: Light Sabre
5/17/2016 | 4:32:00 PM
Re: Time to market and playing catchup
The challenge with all the upgrades is the depreciation cycle.  Telecom used to be a very predictable market with upgrades.  When products were fully depreciated, they were replaced.  Now technology moves faster than finance.  So, people have spent lots of money to recapture their existing customers.  That is a terrible way to run a capital asset business.  The question is to skip right from copper to fiber and there has really only been 1 generation of that.  In residential, I would not expect a new generation of deployment until we get to GPON being congested or an average bandwidth per home of about about 100Mb/s continuous usage (call that 10 HDTV streams for each home 24/7/365).

seven

 
Page 1 / 2   >   >>
Featured Video
Upcoming Live Events
September 17-19, 2019, Dallas, Texas
October 1-2, 2019, New Orleans, Louisiana
October 10, 2019, New York, New York
October 22, 2019, Los Angeles, CA
November 5, 2019, London, England
November 7, 2019, London, UK
November 14, 2019, Maritim Hotel, Berlin
December 3-5, 2019, Vienna, Austria
December 3, 2019, New York, New York
March 16-18, 2020, Embassy Suites, Denver, Colorado
May 18-20, 2020, Irving Convention Center, Dallas, TX
All Upcoming Live Events