G.fast: 1 Gigabit per second DSL

G.fast: 1 Gigabit per second DSL

Summary: DSL, slow old DSL, may yet come back to compete with cable and fibre for the last-mile internet speed championship.


I used to use DSL. It was great in its day, but 6Mbps per second just doesn't cut it these days. So, I turned to cable, where I'm currently enjoying 100Mbps, and dream of Google Fibre's 1 Gigabit per second (Gbps) speeds. However, a new DSL technology, G.fast, may bring old phone lines back into the internet speed race with 1Gbps speeds — even as much as 10Gbps.

G.fast will be competitive with fibre's last-mile internet speeds.


The 1Gbps on copper technology is on its way to becoming a standard. Though there were hopes it would be finalized by now, it now looks like it won't be finished until late this year.

The sponsoring organization, the International Telecommunication Union (ITU), claims that G.fast "combines the best aspects of fibre and ADSL2." By this, the ITU means that consumers will be able to buy a G.fast modem, attach it to their land-line phone connection, and, presto, they'll get 1Gbps speeds. With this kind of speed you'll be ready to watch 4K video.

Today with VDSL2, DSL tops out, practically speaking, at 100Mbps. Far slower speeds are much more common. As the FCC observed in its 2014 report on Consumer Wireline Broadband Performance in the US, "ISPs using DSL technology show little or no improvement in maximum speeds."

This is because, the report says, "DSL, unlike cable and fiber technologies, is strongly dependent upon the length of the copper wire (or "loop") from the residence to the service provider's terminating electronic equipment, such that obtaining higher data speeds would require companies to make significant capital investments across a market area to shorten the copper loops."

Well, that's not going to happen. Indeed some companies, such as Verizon, have been accused of letting its copper network fall into disrepair. If G.fast takes off, those old copper telephone lines will become valuable property again.

G.fast drastically increases speed over copper by using wider frequency profiles than earlier versions of DSL. While VDSL2 uses 17 or 30MHz, G.fast will work on 106MHz and eventually at 212MHz.

To operate at these speeds and bandwidths, crosstalk — interference between adjacent wires — would be a real problem. To take care of it, G.fast uses vector processing to constantly monitor the wire's noise conditions and create an anti-noise signal to cancel it out in a manner similar to the way noise-canceling headphones work.

Still, G.fast isn't a perfect, drop-in replacement for traditional DSL. Its chief problem is it's only capable of delivering 1Gbps to no more than 100 meters. It's being designed to work at up to 250 meters, but it will be slower: 500Mbps. While G.fast is u nlikely to bring 1Gbps to rural areas, it would still work in suburban and urban environments.

Many telecomm equipment vendors, such as Alcatel-Lucent and Huawei are already working on G.fast silicon and modems. The telephone carriers, however, are moving slower. In Europe, Swisscom, Telekom Austria, and German Telekom are testing the technology. AT&T, which has retained its copper network, is believed to be interested in deploying G.fast.

At this time, it appears that G.fast will start being deployed in late 2015 or early 2016. The DSL-based ISPs may want to push that schedule up. Consumer and business want Google Fibre speeds and with the recent approval of DOCSIS 3.1, cable will also soon be able to deploy speeds of 1Gbps and above to home and office.

One way or the other – G.fast, fibre, or DOCSIS 3.1 – we're getting closer to the Federal Communications Commission's (FCC) dream of gigabit internet speeds being available in all US states by 2015. If the DSL-based ISPs want to be part of the conversation, they'll need to move forward with G.fast as quickly as possible.

Related stories:

Topics: Networking, Broadband

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.


Log in or register to join the discussion
  • Strangled by loops.

    I suspect so few will benefit from this development, that it won't take off at all. Even in urban areas, to achieve a 250meter loop - let alone 100m loop - will require absolutely huge investment in local delivery equipment.

    Much more than cable ever cost. So what's the point, except for a very few areas?
    • On top of that

      On top of that Verizon pretty much stopped maintaining their copper pairs in this area years ago and I'm sure we're not alone. Now they just run the fiber to the house and do POTS off that terminal. The downside is the loop current comes from an outlet in your house, so it doesn't work during a power outage. There's a battery in it but that only goes for a few hours.
      Buster Friendly
    • It is actually quite useful

      G.Fast is actually a useful technology where the Telco is committed to FTTP - DSL technologies allow customer premises to be hooked up at fibre equivalent speeds despite limitations preventing last mile drops for whatever reason. For example, G.Fast and XG.Fast will be useful in FTTB scenarios.

      That doesn't mean it is an appropriate technology to provide high speed broadband to an entire country, or even a city or suburb - it is a precise tool that can be used in a targeted way to meet specific objectives where it is not economically, commercially or legally viable to deploy FTTP. But make no mistake - G.Fast is an expensive and complex solution that need a powered node very close to the premises termination point and any nearby last mile copper sharing the same conduit must also be controlled by the G.Fast node in order to eliminate crosstalk - no legacy voice or DSL for your neighbours or it won't work.

      The problem is G.Fast has become a political football for deliberately deceptive politicians like Malcolm Turnbull. He has been claiming G.Fast is capable of greater than 1gbps over 100m, which is patently false - you're talking 70 - 80m for gigabit, even using XG.Fast, and the drop off in performance as you go further out is sharp - by 400m it is completely useless. It's also useless over damaged or lower gauge cable, such as has been used for the majority of Telstra's network. It is a short length solution where you need to get from the street to the building, or for reusing in-building copper for FTTB deployments.

      You *CANNOT* build a broadband network on G.Fast as you would need a node for every three premises in typical suburban areas - it would cost three or four times as much as FTTP.
  • 250m???

    250m? Who the hell lives that close to the exchange? There are people who can't get ADSL2+ right now because they live beyond the 4km limit currently in place. This is just another one of the experiments that works in the lab but has no real world usage, as normal the scientists that develop these technologies are completely out of touch with the real world usage scenarios (or, more likely, the journalists reporting on these stories completely over exaggerate the real world implications of scientific findings.)
    • See above

      See my response above. It is useful, just not as a replacement for ADSL - ADSL is superior beyond 400m than both G.Fast and VDSL2. There are plenty of useful and beneficial scenarios for G.Fast in the real world, just none of those Malcolm Turnbull and his NBN posse are spruiking, because they'd actually need to commit to FTTP for the vast majority of the country before G.Fast to be useful or deployable.

      But Malcolm doesn't let facts or science stand in the way of telecommunications pseudo-science to make his political rhetoric seem reasonable.
  • DSL is still very popular

    DSL still has a significant price advantage over cable competitors. A lot of non-techies don't find the extra speed worth the cost. 6Mbps is fine for most standard web usage, even Netflix streaming as long as you don't have multiple users streaming at once. It's like the tablet revolution - the casual user finds an iPad to be "good enough" even though it is far less powerful than today's desktop PCs. The 1990s days of needing high-end tech is over for a vast majority of people.
  • G.Fast vs GPON

    For cost effectiveness the recommendation was to use G.Fast where FTTH was not economical due to restricted access to get the fibre all the way to the premises. They made no claim that it was cost effective as an alternative where FTTH was no suitable due to political or ideological reasons, no reasons of the Communications Minister's ego.
  • Horse Balogna...

    I read a similar article that claimed this new tech was at best only beneficial for FTTC deployments. In other words, it might work great to bring fast speeds into a home from the curb or a very nearby node.

    As to why ANYONE would want to leverage copy rather than consider a last 500 feet wireless solution is beyond me. My local cable company already runs new drops from the pedestal to the home anytime someone complains about their digital picture's stuttering or their HSI getting cranky. If coax has age problems, I can only imagine the problems faced by even older copper.

    Wouldn't a MIMO wireless solution from the curb to the home be a much better solution and in the end, be much more reliable as there's no last 500 feet of questionable copper?
    • Spectrum

      Because wireless signals are a shared medium that have finite spectrum availability and are subject to variable interference. It's also comparatively expensive, but unless you're willing to provision a similar number of base stations as you would need G.Fast nodes there isn't much point. You have the same electricity delivery problems, they will actually draw more power than micro-nodes, they will have to be individually calibrated to not interfere with their neighbours and they will be subject to lower performance with bad weather, solar events and things like trucks getting in the way.

      Long-range wireless makes sense when the geography and population density prohibit cost-effective fixed line delivery, and technologies like mobile broadband are useful for mobility, but fixed line broadband are far cheaper and more reliable solutions for the vast majority of Australian premises.
    • robradina .. Alternative solution? Sure, "better solution"? Not so fast

      Well, that's just the thing. Wireless to your door, as opposed to DSL or FTTP, has its own issues / problems. For starters, the cost to deploy, secure and maintain curb-side wireless creates not so obvious costs and security issues. I won't go into minutiae, but if you haven't considered the pitfalls of wireless to your door, you might wanna read up on them (hint: Google it ;)

      Besides, it's simply one solution amongst a few feasible solutions.. that's all.

      But i do agree, trying to trump up old copper to perform a task it wasn't ever designed for originally - especially beyond a certain distance over a decrpit, old POTS line, seems a completely hair-brained idea. I personally think it's a gimmick thought up by some marketing guy who happened to be able to sell it to a few I.T guys. They obviously saw $$$ signs at the thought of somehow "reanimating" old infrastructure between exchanges *without* ever doing a Cost-Benefit analysis (or TCO) for the idea.

      In plain terms, G.Fast is a nice idea on paper.. but would easily be a freaking nightmare, ongoing costs-wise (e.g. line maintenance, etc), to get up & running ... let alone maintain over the long term.
      • Line maintenance a red herring

        This is the problem with this topic - lots of opinions laden with mistaken and misunderstood arguments masquerading as facts.

        Over 70m there would be pretty much no maintenance issues for copper lines qualified as being in good condition and suitable for G.Fast, bearing in mind that would only be a tiny proportion of existing CAN customers in Australia in the first place. Failing someone severing the copper line with a shovel it should be good for 10 to 20 years. A great deal longer for in-building copper where G.Fast to deliver FTTB is considered suitable.

        That doesn't mean G.Fast isn't much more than a misleading distraction in the broader telecommunications infrastructure debate - see above for reasons why you can't build an network using it.
        • Oops

          People need to stick to facts they actually understand, that are actually applicable, and stop muddying the waters with baseless assertions of fiction in an attempt to shore up their argument - the arguments against FTTN as the national infrastructure solution stand up very well on their own without adding confusing BS, thanks.
  • Line Length?

    So, how many residences will be within 300 metres of an exchange/node? This sounds all very nice, but is totally ridiculous when there is a FAR better technology already being installed in Oz - FTTP! You can get Gigabit Broadband anywhere up to 15-20 km's *guaranteed*, without needing a node on every corner.

    The other "Elephant In The Room", of course, is that all these tests are great on a piece of nice new copper. Put it into the ground and open the cable up every 30 metres and leave the joints in sweaty pits and it goes bad pretty quickly...

    Why are we even having this discussion anyway?
    Fat Pat
  • G.fast standards were confirmed by ITU in April

    A lot of people have been under the impression standards have not been completed but they are wrong. This article indicates commercial launch by 2016 but several articles have indicated the race to bring OEM to market will move launches into 2015 because final standards were approved by ITU in April'14 versus end of year 2014. OEMs now have ears pinned back in a race to launch product. Eight month time frame shifted forward.

    People here bagging on G.fast just need to do a little homework. 80% of the cost of installing FTTH is from the curb into the home. Pulling fiber into the home as well as dispatching a tech to meet with a homeowner to gain access into each home and complete the install is the bulk of fiber deployment cost. If the telecoms can run a fiber trunk through the alley, plunk in a node, and drop a G.fast box on the homeowners porch for 20% of the cost of full fiber then this is a massive cost savings. This is big for copper operators trying to stay in the ISP game without the massive amount of capex needed to upgrade to an extensive full fiber network. G.fast allows for a much more rapid mass deployment of 1 gbps service by eliminating the burden of entering each customer's building while removing a large amount of the capex requirement. This keeps them in front of encroaching fiber ISP providers. As those individual copper lines slowly deteriorate over the next decade or two the telecom can then return to a given premise and install full fiber to the home at their discretion and a more palatable lump of capex. G.fast can't come quick enough for the likes of AT&T who is getting hit from all directions by new full fiber ISPs entering/expanding in its markets. Read somewhere that T converting its entire network to G.fast would be equal to three years capex. Now think of the cost of full fiber at 5X the cost of G.fast deployment. That is exactly why G.fast is being discussed. Keep in mind T's capex has included mobile services infrastructure. G.fast cost savings is huge.

    Fat Pat, there are real world tests already pushing across opened copper at the node - multiple field trials. The nice benefit is you can push up to 16 customers through each node. Key is how many customers can you strategically locate a node within 100m. Varying reports but appears AT&T has fiber nodes at an average of 2000-3000' from all customers so it will take strategic node placements into a given city block to cover multiple home connections. Fiber back bone is MOSTLY strung but extending 1 gbps is vastly cheaper using G.fast vs. full fiber. Great reason for discussion. Gigabit broadband at 15 kilometers? Would love to be enlightened on this technology.

    P.S. Bell labs playing with higher copper frequencies have shown higher speed potentials over G.fast.