opinion


DSL acceleration: Simple concept, complex reality

Mobile broadband traffic is driven by dongles and embedded laptops

On the face of it, DSL acceleration technologies seem to offer a neat solution to a key problem of fixed line operators: to offer superfast BB without having to invest billions rolling fiber all the way to the home. But the new DSL technologies, as discussed in last week’s DSL Acceleration conference, are not without their own complexities.

Vectoring will not be without its challenges

Vectoring promises to allow operators to boost the speed and reach of their copper-based broadband networks by eliminating a kind of interference between VDSL2 lines, known as “cross talk”. But eliminating this cross talk leads to further challenges. For example, once the cross talk has been eliminated further impulse noises will be revealed that were previously masked by the crosstalk. About 85% of the revealed noises will be from within the subscriber’s home. This will necessitate improvements in so-called dynamic spectrum management (DSM) Level 1 technology (also known as dynamic line management) which aims to keep lines stable and without resynchs that lead to customer downtime and irate calls to customer service centres.

Although ADSL from the central office and VDSL2 from the cabinet would not cause terminal problems for vectoring, with ZTE claiming losses from optimal vectoring gains of around 10% thanks to their use of upstream back off power, there will be problems with VDSL2 from the incumbent and sub loop unbundling. Many of the delegates seemed sceptical that sub loop unbundling requirements would be lifted, although this has already occurred in Belgium.

In addition operators seemed sceptical on the realism of deploying vectoring in the very short term because of the lack of commercial rollouts. Availability of chipsets will need to improve with Ikanos currently refining their vectoring technology. Alcatel Lucent will use its own chipsets but will need to perform more field trials for its first commercial vectoring customer, Belgacom, before the operator can launch.

There will also be elements of vendor lock-in with the technology. For example, an operator need only replace its VDSL2 line cards in order to support vectoring – provided the cards are from the same vendor of the DSLAM. In order to use a rival vendors’ vectoring technology, the operator would have to replace the whole DSLAM including shelves and cabling. To some extent this could increase prices as there would not be a level playing field between competing vendors. Interoperability between different vendors’ chipsets on the DSLAM and customer premise equipment (CPE) sides will also not be available initially, again increasing the risks of vendor lock-in.

Despite these constraints I am still of the opinion that vectoring is a very promising technology and one that operators would be wise to examine very carefully.

Pair bonding looks set for more deployments in the next few months

Many of the issues described for vectoring do not apply to another form of DSL acceleration technology, VDSL2 pair bonding, which involves bonding two or more copper pairs to increase speed or reach. Chipsets are on the market and there are already operator customers for VDSL2 pair bonding. Back at the Broadband World Forum 2011 conference in September, Alcatel Lucent claimed to have about 10 commercial deployments and Zyxel was aiming to sell a combined solution of VDSL2 bonding-capable line card and CPE which would help cut down on integration work and appeal to smaller regional incumbents. Availability of VDSL2 bonding CPEs was an issue in the past but as vendors such as Zyxel and Comtrend move into this space prices will fall.

Obviously, the big drawback for pair bonding is the availability of  households connected to two or more copper pairs, but this seems less and less of a problem. The availability of two pairs per household would appear commonplace in Malaysia, for example. About 40% of homes in Poland are also passed by two pairs which would enable Polish incumbent TPSA to increase the coverage of its mid-tier offering of 25-40Mbps by around 16%.Canadian incumbent Telus also described pair bonding as compelling and has already deployed the technology in Quebec.

Of course, two pairs are not available everywhere. For example, in those markets, such as France, where ISDN lines were widely deployed in order to enable two simultaneous calls to be made over one physical line, two pairs are largely not available.

Pair bonding and vectoring are only two of the copper-based cards operators can play

One of the other features of the conference was the sheer range of technologies that operators can use for copper both now and in the future and that extends well beyond vectoring and bonding. For example, operators can move from using 17MHz of spectrum for VDSL2 to 30MHz of spectrum in order to increase speeds offered to end customers. However, many regarded 30MHz for VDSL2 as an immature technology, which means additional cost and complexity at the chipset level and has yet to benefit from economies of scale as it has yet to be deployed in large volumes. There was also talk of various technologies to improve line stability such as the aforementioned DSM Level 1 and G.INP, which has already been standardised and is available in commercial chipsets. Further technologies will also emerge that will build on the first wave of technologies to prolong the life of copper, for example, phantom DSL which first requires both pair bonding and vectoring.

Copper technologies will continue to mature and evolve

This plethora of technologies, with the promise of further enhancements to come with the likes of G.Fast, means that interest in copper will remain for the foreseeable future and commercial deployments of technologies such as VDSL2 pair bonding and vectoring will continue to grow.

Tags: ,

Leave a comment

Your email address will not be published. Required fields are marked *

Events

There are no upcoming events.

Polls

What is your name?

Loading ... Loading ...