How viable is wavelength unbundling on FTTH networks?

I have been looking into the Open Lambda Initiative (OLI), NGPON 2 technologies and the fascinating framework that the OLI is creating which would allow multiple operators to each deploy their own active equipment on a PON tree architecture and use different wavelengths on the fibre. This offers a solution to a problem with today’s EPON/GPON FTTH  systems: only one company can deploy its own OLTs, limiting flexibility and room for differentiation amongst the other players.

The idea of the OLI would be to have a neutral spectrum management authority which could, for example, ensure that different wavelengths would not interfere with each other. What different models could emerge from such a framework, however? Here we look at the different possibilities, which are the hot ideas and which seem unrealistic.

Third party financial investors and utilities

A financial investor might be attracted to building a PON FTTH infrastructure because the costs of rollout might be 30 per cent less than for a point to point network. With a model where each operator on the network could deploy its own active equipment the risk of overbuild would be reduced giving greater security to the investor. Personally I do not find this argument too convincing. If we look in Western Europe pretty often we will find cable networks, often performing very well in terms of market share thanks to their Docsis 3.0 upgraded networks. The same applies to the USA and Canada. These cable operators are going nowhere: any financial investor looking for a business model where there are no competing infrastructures better look elsewhere.

What about the less developed markets where cable might not be present? But in less developed markets rollout costs of FTTH tend to be lower, for example, due to the higher preponderance of MDUs or the ability to deploy overhead cables. Bottom line: those thinking about becoming active in FTTH in these kinds of markets might as well build their own network- why share the revenues with a third party infrastructure provider?

Utilities investing in FTTH networks, for example because they have existing ducts, might also be attracted by the OLI framework. The fact they have existing ducts makes the business case for them much easier than for third party financial investors who may be more worried about overbuild. Often utility investors are not interested in managing active equipment and so the OLI framework could also be positive in that sense.


One argument put forward is that with wavelength unbundling incumbents can press ahead with FTTH investments safe in the knowledge that drastic regulatory remedies such as functional separation will not be applied. But again this seems unconvincing. Certainly in Western Europe incumbents have been reluctant to roll out FTTH and rightly or wrongly one of the reasons for this is the fear that they will be forced to open their networks. The fears the incumbents have, whether correct or not, are real. Allowing wavelength unbundling over a PON tree architecture would likely only make this situation worse, creating a disincentive to invest for the incumbents.

Linked to this argument is the idea that an OLI environment would potentially allow today’s OTT players access to individual wavelengths. But introducing a new competitor in this way would again likely discourage incumbents from making investments in the physical fibre infrastructure in the first place.

Of course in Western Europe FTTH rollouts remain limited but the situation is very different in Japan, indeed DSL unbundler Softbank and NTT are both members of the OLI. So in theory since the FTTH network already exists it might be easier to move towards wavelength unbundling in Japan. Still the regulatory will needs to be present and an additional complexity might be added if NTT deploys 10 GEPON. This would leave the incumbent with EPON, 10 GEPON and RF video systems. NGPON 2 technologies which make wavelength unbundling possible could coexist with the 3 other technologies but a potential lack of flexibility could be incurred.

In other markets, for example in Latin America, DSL LLU is not available and so the regulator imposing wavelength unbundling on incumbent built FTTH networks seems unlikely.

Co-investment scenarios

Allowing wavelength unbundling  might also make co-investment in FTTH networks easier. Without wavelength unbundling if the operators deploy a GPON infrastructure then only one party will be able to operate the active equipment. The players could deploy a point to point infrastructure but this would push up costs, making an already problematic business case even more difficult. Of course part of the rationale for co-investment, for example in a market like Germany, is that competition from cable means that both incumbent and DSL unbundlers need to do something to arrest their decline. But if there is cable then the argument about financial investors not wanting competing infrastructures starts to look hollow.

Incumbents might also view the advancements in VDSL technologies, such as vectoring, as an opportunity to lock out DSL unbundlers and move them back to bitstream based offers, no matter how advanced the virtual unbundling offers are in that particular market. Whereas incumbents might see co-investment as a necessity for FTTH, the much lower costs of FTTC may mean that incumbents are prepared to invest in this technology alone. So a mixed picture on co-investment, which in any case can be achieved through different means other than wavelength unbundling. For example, if co-investment is defined in the broadest sense, in very densely populated areas in France shared infrastructure can be deployed within an apartment building but there will be multiple networks with cost oriented access to ducts beyond a flexibility point.

Government backed open access networks

This looks like a hit. Governments, such as in Australia with its NBN Co, see rolling out FTTH as a way of developing their economies. The governments are wary of becoming involved as retailers of broadband access to the end customer so may become the operator of the active equipment on the GPON network or delegate this to a third party. But this leaves the ISPs with less flexibility to deploy their own services, particularly if the bitstream service from the network operator is only based on layer 3 access. One way out of this problem would be a wavelength unbundling model: the government rolls out the infrastructure and then leaves the ISPs to deploy their own equipment and differentiate themselves.

There are also markets where the incumbent is not active as an ISP for example, in Kuwait or with a Gauguinesque trip down to the South Pacific in New Caledonia. An OLI framework to allow wavelength unbundling would allow the incumbent to upgrade its copper infrastructure to FTTH, potentially benefitting the economies of these countries, and allowing the ISPs flexibility in deploying their own active equipment.


The best bet for an OLI type model is probably from government backed open access networks. Co-investment models could also emerge in the future and the model could also be attractive for utilities. It is worth bearing in mind, however, that NGPON 2, a technology which would allow different operators to use different wavelengths, will only become commercially available in 2015.

Informa TM will shortly publish more research on some practical issues surrounding multiple operators using different wavelengths on a PON tree infrastructure. For example, how would spectrum be allocated- via auction at the start or on an ongoing basis?? How dynamic would the release of spectrum be? Who would manage the spectral assignment agency? The Open Lambda Initiative seeks to define a mechanism by which various models, for example, whether spectrum is allocated by auction at the start or on an ongoing basis can become a practical reality.

Tags: , ,
  • Cable Next-Gen Europe Digital Symposium

Leave a comment

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

This site uses Akismet to reduce spam. Learn how your comment data is processed.