G.fast Plugs the Gap for Gigabit Cities

The race to deliver gigabit broadband services is accelerating, and G.fast is the technology that telecom service providers need to compete successfully. G.fast will change the market for high-speed broadband and could reverse the subscriber migration to fiber and cable services.

G.fast builds on the success of VDSL2 and removes significant limitations. The first chipsets to support G.fast are already available, the first subscribers have been connected and many service providers are running trials that demonstrate performance up to 1 Gbit/s.

Digital subscriber line (DSL) technologies, including ADSL2+ and VDSL2, have enabled telecom service providers to extend data bandwidth over existing copper telephone lines from 50 kbit/s using dial-up services, to 100 Mbit/s or more using VDSL2 with vectoring and bonding. The increased data rates are only supported over short copper lines, so service providers are reconfiguring their infrastructure to reduce typical line lengths. Service providers can dramatically increase bandwidth without the cost of installing fiber to the home (FTTH) by installing fiber to a distribution point (FTTdp) close to the subscriber. This approach is key for G.fast, which is most effective over copper lines shorter than 200 meters in length.

Chipset vendors have continued to develop VDSL2 chipsets, expanding support for vectoring and bonding. Vectoring uses DSP technology to apply Far-End Cross-Talk (FEXT) cancellation across multiple lines. This dramatically increases the real-world performance of VDSL2 closer to the theoretical maximum and is a key part of G.fast. VDSL2 bandwidth can also be increased by bonding multiple lines. The first G.fast chipsets have been developed by taking existing VDSL2 solutions, with minor changes, and updating the software to support G.fast or as a new development.

At this critical time, the latest edition of Heavy Reading Components Insider, "G.fast Turbocharges Next-Generation Access," analyzes G.fast, VDSL2 and ADSL2+ chipset solutions. The report covers the leading chipsets, reviewing product features, performance and flexibility. The report profiles the four leading vendors, examining their strategies and product portfolios.

Central office and distribution point chipsets are available from Broadcom, Ikanos, Lantiq and Sckipio. The chipsets support between four and 36 ports and consist of separate digital front-end, analog front-end and line driver devices. The Broadcom chipsets support G.fast, VDSL2 and ADLS2+ or VDSL2/ADSL2+ with, or without, vectoring. Chipsets from Ikanos and Lantiq support VDSL2 and/or ADSL2+. Both companies also have VDSL2 chipsets specifically designed for distribution point applications. Sckipio is focused on G.fast and is sampling a complete chipset for central office and distribution point applications.

Customer premises equipment (CPE) chipsets are also available from all four companies. Broadcom and Sckipio are sampling chipsets that support G.fast. Marvell and Ikanos are developing chipsets that will support G.fast. Chipsets from Broadcom, Ikanos and Lantiq integrate a gateway processor to support a range of applications, including residential and small-office routers and voice/wireless gateways. Some devices integrate full G.fast, VDSL2 and/or ADSL2+ transceivers. Others integrate just the data pump, or nothing. Lantiq and Sckipio have developed a G.fast residential gateway reference design that uses the Lantiq gateway processor and Sckipio G.fast transceivers. Intel has recently agreed to acquire Lantiq, expanding the company's business in the broadband and smart home market. (See Intel Targets 'Smart Home' With Lantiq Acquisition.)

Broadband service providers need to offer a full range of packages from basic broadband through 100Mbit/s services to 1 Gbit/s and above. The combination of G.fast and VDSL2 provides telecom service providers with a solution that supports gigabit broadband without installing FTTH. This allows them to continue benefiting from their installed copper infrastructure and offer cost-effective broadband services that can compete with services from fiber and cable companies.

— Simon Stanley, Analyst, Heavy Reading Components Insider

This report, G.fast Turbocharges Next-Generation Access, is available for $595. For more information, please visit: www.heavyreading.com/commchip.

brooks7 2/18/2015 | 10:14:33 AM
Re: G.Fast t,

You can call it FTTdp but it is just FTTC.  Let's go ask DT how they liked deploying FTTC and then having to rip it out of the ground.

But beyond that, no I think there will be a few spots where G.fast gets deployed.  But let me ask you, didn't we have the same hype around VDSL2?  Why yes we did.  And very little of it got deployed in Europe.

Asia will see some deployment in high rises but NTT and KT have already shown that FTTH is doable in Asia.

The problem with short loop DSL remains the same.  It takes 90% of the fiber construction of FTTH to deploy and then you still have a copper loop and powered OSP.  If you want to take a Euro-Centric view, then think about this....what percentage of folks are on street cabinets?  The answer is very low - generally 10% or less in Europe.  Now go make that 90% and go see how the model changes from an operating cost standpoint.


t.bogataj 2/18/2015 | 2:57:22 AM
Re: G.Fast @seven

In short: You are dismissing FTTdp and G.fast because you cannot see beyond the narrow US-centric perspective.

brooks7 2/17/2015 | 8:31:33 PM
Re: G.Fast Jason,

They are using RUS and CAF to gold plate networks in the IOCs as they have for decades. G.fast won't go there except maybe off of some apartment buildings. The RBOCs are abandoning these lines so no business there. Hong Kong and similar spots...maybe. Europe is the primary spot. Competition is a lot less and they can do a deployment and pretend to be forward looking. Let us compare for real. Go add up all the homes in Euope on FTTH and compare that to the number of homes on FiOS. Then let's talk about networks that are behind.


Edit P/S: Let's remember something here.  In all our physical communications, we are still talking about the propogation of E-M waves in a medium and that the information theory of Shannon and Nyquist still provide boundaries to what is possible.  G.fast is doing some clever ways of bumping the Signal-to-Noise Ratio (SNR) on these copper pairs.  But this game only goes so far.  

At the end of the day with copper, bandwidth is limited by distance.  If you want high rates, you need short loops.  To get short loops, you need to roll out more fiber.  When you roll out short loops the number of units served by a DSLAM go down dramatically.  We were working on 96 port line cards in DSLAMs for ADSL and ADSL2+.  Can you imagine a situation in which you could deploy and use all the ports on a 96 line DSLAM with G.fast (in other words 96 terminations with a radius of 500 feet)?  This means that the cost of commons goes up rather dramatically and thus price per port.

Next up, you have to power the things.  Unless you want to depend on having enough take rate at all DSLAMs to power it, you need to bring power out to the edge.  Ask folks how hard it is to do FTTC powering from upstream elements.  If you don't do that, then you need to have AC dropped to about every dozen or so homes to power the units.  If you drop AC there, you need to have POTS bypass or you will have batteries there.  Maintenance of that many batteries will kill the ongoing costs.

So, I think there are some high rise and similar MDU places that G.fast can go.  But in SFH construction, there are a lot of issues.

jasonmeyers 2/17/2015 | 8:23:00 PM
Re: G.Fast I'm curious about all the rural carriers that are deploying fiber though -- if they're already deploying fiber drops to homes to deliver gigabit speeds, in many cases, they must be able to make the economics of fiber work (right?). So why would they even look at something like G.Fast? Different scenario? 
brooks7 2/17/2015 | 8:08:53 PM
Re: G.Fast Jason, I am trying to,tell you that there are lots,of places where FTTH today. Rural,is an example. No savings in 1 home per DSLAM. Remember there is a huge fiber construction project to make g.fast work. Have to shorten loops from either 12 Kft or 2.5 Kft to 500 ft loops. Plus you have to drop a boatload of power drops. Bellsouth liked FTTC as did what is now Centurylink in Vegas. The rest of the carriers did not do it. Even then, BLS could theoretically get 100 Mbps out of FTTC 15 years ago. Could talk to Bill Smith about it, last I looked he was still around. seven
jasonmeyers 2/17/2015 | 6:56:21 PM
Re: G.Fast "In the long term, fiber will be the vehicle. "

Seven - are you saying that the cost of deploying FTTH has dropped to the point that it's just as economical to deploy as it would be to deploy G.FAST to upgrade the copper to be able to achieve comparable speeds? 
brooks7 2/17/2015 | 12:01:11 PM
Re: G.Fast I think the question for G.FAST is where the money saving is that makes people want to deploy it?

To get to the 1 Gb/s speed in a single family home context we are talking high density surburbia or urban settings.  In that case, fiber is brought to the cross connect and then the drop cable remains copper.  This also means power is brought to that local box unless that crazy backpowering scheme is going to be attempted.  (What happens if you don't have a big take rate on the service?)

In a multi-dwelling context, it depends on size.  In smaller buildings, you could live with the DSLAM in the basement approach.  But in a larger building, you may have to go to at least a DSLAM per floor.

So, in these cases the savings is that last 100m or so of cabling.  The challenge is that you still have to do a lot of fiber construction to get to having those fiber fed mini-remotes.  On top of that, you have the powering issues to resolve.  I would also like to think about how good the splitters have to be to allow for POTS bypass and ensure the high data rates.

All of this has led me to the conclusion that the G.Fast Market is Smaller than the VDSL Market which was smaller than the ADSL market overall.  In the long term, fiber will be the vehicle.  The question is how much does someone save over what period of time to do G.Fast.


davidhoffman5 2/17/2015 | 11:43:43 AM
G.Fast G.Fast sounds wonderful. The problem is that AT&T and other USofA legacy telephone companies failed to deliver the full capability of ADSL2+ to each potential subscriber. If the telephone companies had invested in the remote fiber connected DSLAMS that were needed to do so, all DSL customers would be getting 20Mbps download capability today. They are not getting that level of service. As far as VRADs and VDSL, that has an even smaller geographic footprint in the USA. Maybe some other countries will benefit from G.Fast, but I doubt the USA will. 
Sign In