Are sub loop unbundling and vectoring incompatible asks Stephen Wilson

I have been conducting some rigorous Philip Marlowesque investigations into an important topic within the fixed broadband access family, operator coexistence with vectored VDSL2, and more specifically whether Dynamic Spectrum Management (DSM) can allow vectored VDSL2 to coexist in the presence of sub loop unbundled lines. The idea that vectoring and sub loop unbundling are incompatible is that there will be alien crosstalk from the lines of the sub loop unbundler that will cause significant losses in the gains that vectoring can provide.

If coexistence between vectored VDSL2 and sub loop unbundling is not possible this would mean that if an incumbent (or alternative operator) were to roll out FTTC and deploy vectored VDSL2 then it would not be possible for an alternative operator (or incumbent) to roll out fibre to the same cabinet and deploy its own DSLAMs. But a debate is emerging within the industry about possible scenarios for operator coexistence when vectoring is deployed. Let’s try to examine with some economy the different scenarios that are possible.

How might DSM allow sub loop unbundling and vectoring to coexist?

DSM can control the transmit spectrum of different lines, sometimes in coordinated fashion, using a spectrum management centre (SMC). DSM can dynamically shape the power spectral density of lines or how the power on a line is distributed across different frequencies. By doing this it may even be possible to reduce the power on a line and have higher bit rates because crosstalk has been lessened. In a sub loop unbundled environment the idea would be to lessen the alien crosstalk that reduces the performance on the vectored lines.

In a sub loop unbundling environment one potential application of DSM might be for each operator to have its own SMC and certain rules to be established between the SMCs to ensure optimal performance across the lines. The regulator could set these rules, which would then merely be implemented by the operators, to establish a maximum benefit across all the lines, although formulating such rules would require some negotiation between operators. Another scenario would be if the different SMCs could communicate with each other, which is technically possible. It is important to stress that it is not necessarily the case that only one SMC is deployed between the 2 operators. Such a scenario is plausible but would lead to inevitable problems of who actually controls the SMC.

So there are possible ways in which operators can potentially use DSM and help maintain the benefits from vectoring. But let’s examine whether these are of practical use in the field.

Scenario 1: vectored VDSL2 with plain non vectored VDSL2 from another operator

Software vendor ASSIA reports potential limits in losses of only 5-10% on vectored lines when DSM is used on non vectored sub loop unbundled VDSL2 lines. But there are 2 important assumptions used to generate this result (a) all loops are of equal length, (b) bit rates on the non vectored lines are capped at 45Mbps. Let’s examine how realistic these assumptions are.

Equal loop lengths is obviously not exactly the case as a cursory look at the position of a cabinet on a street can attest. Is it really such a bad assumption, however? For example, in the UK access network around 45% of lines are between 200 and 500m from the street cabinet. Nevertheless every access network is different so it is difficult to draw more general conclusions about how equal loop lengths are in other countries.

The second assumption of capping the bit rates on the non vectored lines at 45Mbps also needs to be examined. If the bit rates on the non vectored lines are capped at this level then the vectored lines can deliver 100Mbps downstream to 350m for 99% of subscribers. How realistic is it that a sub loop unbundler cap its lines at this level? In terms of pure bit rates 45Mbps would still give a good performance for the alternative operator and does not seem like a terminal constraint. To me it seems like a more favourable scenario than for an alternative operator that has deployed VDSL2 from the central office being forced to desist from using this technology to make way for vectoring, as has happened in Austria.

Returning to the original point if the sub loop unbundler implements DSM level 2 to help the vectoring operator he himself would also get additional benefits, for example, in terms of OPEX savings through power reduction. But how easy would any commercial negotiation be in determining the trade offs in terms of the bit rate capping that would be necessary on the non vectored lines?

One example to demonstrate the potential difficulties would be where the sub loop unbundler has only one line nestled in between a number of vectored lines from the incumbent. If DSM were deployed by both operators the incumbent would see far greater benefit than the sub loop unbundler because it has more lines and also because the single line would be causing a large amount of crosstalk into a number of the incumbent’s lines. This could make commercial negotiation tricky. An alternative point of view is that such commercial negotiation would not be that difficult as the situation about line distribution in one cable would likely be balanced out by the situation in another cable.

There is also a link between loop length and a bit rate cap of 45Mbps. For loops above 1000m (which in the UK case is around 10%), assuming that the incumbent rolls out fibre to the cabinets serving these loops, vectoring may not be used on all the lines because the loop length is prohibitively long (vectoring works better on shorter loops). Instead the incumbent might offer plain VDSL on the longer loops but this itself may generate significant crosstalk because these lines may need to operate at full power to guarantee a minimum service level. But by operating at full power, power reductions may not be possible to limit the crosstalk onto the sub loop unbundled lines and a bit rate of 45Mbps on these lines may not be possible.

Do I consider that a scenario where the sub loop unbundler has deployed plain VDSL2 and the incumbent has deployed vectored VDSL2 to be implausible? It might be unlikely, but it is not impossible. Probably the largest sub loop unbundler in the UK is the Digital Region Project in South Yorkshire, but it seems there is a shortage of funds, which might make it difficult for an upgrade to vectored VDSL2. One could envisage a scenario where the incumbent rolls out vectored VDSL2 in the same area and in this way coexistence probably would be possible, at least initially, but this leads us to our next scenario….

Scenario 2: two operators both having deployed vectored VDSL2

Technically this scenario is not really any different to scenario number 1, the problem is still that the alien cross talk, this time from the other vectoring group, needs to be cancelled. The fact that the other lines are vectored does not make any real technical alteration, the only slight change is the slight increase in power, of the order of only 1%, associated with generating the anti interference to cancel the cross talk.  But whilst technically the situation is the same the 2 scenarios are different in terms of bit rate capping.

Let’s examine the situation more closely. Alcatel Lucent has used a model where there are 2 groups of 4 vectored lines of length 400, 600, 800 and 1000m. By using a DSM technique called optimal spectrum balancing the vendor believes that there would be limited gains on the lines. Again one set of lines could be capped at 45Mbps and that would allow around 90Mbps to be delivered over the other vector group. But why would an operator deploy vectoring and then cap its bit rates to a level that could be provided with plain VDSL2? In practice if the first operator is allowed to continue to offer around 90Mbps then no second operator would ever really have a motivation to roll out vectored VDSL2.

There is another scenario, however, where both operators cap their bit rates at a relatively modest 60Mbps, which would give both players a boost in bandwidth over what plain VDSL2 could provide. It might be that the operators consider this boost in bandwidth sufficient to justify the investment in vectoring, on the other hand it might not.

But here we return to our earlier example, if we have vectored VDSL2 and non vectored plain VDSL2 then the vectoring operator would be able in theory to offer around 90Mbps. But in practice the non vectoring operator at some point in the future will likely want to upgrade to vectored VDSL2. The non vectoring operator has already made concessions to the vectoring operator in the form of capping his bit rates so would likely want a situation where the vectoring operator makes some concessions to him so he could offer 60Mbps.

If this were to happen however then the first operator to deploy vectoring would have to reduce its maximum speeds from 90 to 60Mbps. This does not seem satisfactory and in practice in the situation of vectored and non vectored VDSL2 the vectoring operator could only really offer 60Mbps because at some point in the future he may no longer have the possibility to offer 90Mbps. Of course there is yet another scenario where both operators decide to deploy vectoring at the same time, in which case they could avoid this problem and have the possibility to offer 60Mbps.

Bit rate consistency and unpredictability with vectoring and sub loop unbundling

One of the benefits of vectoring is also the way in which it can improve the consistency in performance between lines. It is true to an extent that the presence of sub loop unbundled lines in the binder will reduce this improvement in consistency and hence the potential to increase marketable bit rates that vectoring can provide. But even with vectoring and no sub loop unbundling there are still significant sources of noise that have different effects on different lines, for example impulse noise from within the subscriber’s home which is an area where DSL management software can also help.

Summary

Coexistence between vectored VDSL2 and sub loop unbundling is possible but there are big caveats with this statement. Bit rate capping would be necessary and this could make it practically difficult for operators to agree amongst themselves on the best implementation scenario. In practice it would be difficult for any operator to offer above 60Mbps on vectored lines whether the sub loop unbundled lines are vectored or non vectored or even if at that moment there is actually no sub loop unbundling. So basically a regulator could ban sub loop unbundling and this would allow an incumbent to offer the full vectoring benefit or it could keep sub loop unbundling which in practice might mean capping bit rates to around 60Mbps.

More generally speaking sub loop unbundling is highly unlikely to be widely deployed outside of certain regional pockets. All the more reason then, as I have argued before, for all parties to encourage the development of better virtual unbundling offers, as vectoring should not become an excuse for the incumbent to rewind competition.

If any regulators are keen to discuss more, please do not hesitate to contact me.