There is a real issue with fibre-to-the-premise projects that the level of investment and service opportunity is such that it makes almost no commercial sense to compete in the first mile. With an effective natural monopoly, especially in rural areas, it is essential that forms of competition are enforced and the monopoly regulated. With the market for next generation services beginning to emerge in the UK, now is the time to ensure the right regulatory models are in place - before the build-out begins in earnest.

This short piece looks at what models are necessary to ensure contested and open next Generation networks.  

Models for competition

There are four key levels at which competition can be encouraged and for a truly open market, the more levels of competition the better for both subscribers individually and for the wider economy. Those levels are:

• Network element - a dark fibre
• Optical - a reserved wavelength
• Data-link - a reserved PPP tunnel
• Physical - a reserved VLAN

Each of these layers has its own characteristics, which in some case may make then uneconomic to offer today.

Network Element

Offering dark fibres straight from the frame at the central office permits service providers to offer both their choice of services but also the access technology, CPE, and speed - it generates technological competition in the first mile, preventing a situation occurring as it has has with copper in the first mile where the incumbent dictates the technology, making decisions based on retaining market share at the lowest investment level.

The one issue with this model is that it gives a supplier exclusive access to a subscriber; once a dark fibre has been rented to a service provider, no other provider can access it, removing an possibility of separating triple-play offerings for example.

Optical unbundling

In the same way that copper frequencies have been unbundled to permit ADSL competition at the exchange, optical frequencies can, conceptually at least, be unbundled. Using simple, passive splitters at the customer-end and couplers centrally, or using wave-division multiplexors would permit multiple wavelengths to be delivered to subscribers, in the same way that a splitter is used today to separate voice and ADSL signals.

For example, an ISP and a cable TV company could offer services to a single subscriber using their own dedicated wavelengths, with each getting exclusive access to the bandwidth; the ISP could offer 1 Gbps Ethernet Internet access on one wavelength, and the CATV company a full multiplex of high-definition video services.

One interesting conceptual model is one where all or a subset of wavelengths might be handed over to the subscriber, giving the customer the right to allocate their channels to whom their like. This is the time a monopoly provider might conceivably be permitted to sell directly to the end-user; a cost-plus service, handing over two or more wavelengths to the end-user, still permits competition in both the service offering and technology elements as, theoretically, two technologies, such as Ethernet and ATM, might co-exist. With multiplexed wavelengths it may permit some competition in the first mile technology, although without universal access to the entire spectrum it may reduce the options.

Virtual LAN's

On an Ethernet switching fabric, for example, its possible to allocate a dedicated virtual LAN to a service provider. This gives the provider simple access to what appears to them as their own first mile Ethernet network. While the cost of providing a VLAN is significantly less than, say, an optical wavelength, there are some possible drawbacks. A customer may be free to use VLAN trunking services, permitting them to take multiple services from multiple providers. However. the performance is limited to the port speed - two providers vying for a single subscriber will share the port bandwidth, albeit with no sight of each others channel content.

This model permits the service provider to select the optimum middle-mile network for their commercial offering, without investing in first mile technology. However, the reduced investment is likely to attract more competition.

While the VLAN dictates the physical layer - Ethernet for example - there is still a opportunity, theoretical at least, for this kind of competition to co-exist with optical unbundling. The fibre owner might offer a VLAN service as a kind of semi-managed offering on one set of wavelengths, while still reselling other wavelengths to other providers.

IP Access

This is much more akin to today's networks, with a L2P/PPP service offered to service providers. This offers the least level of investment for a service provider, generating the greatest number of competitors - albeit ones with no investment in the future shape of the network. In effect this is as close to a total outsourced ISP service as its reasonable to get.

The fibre owner may chose to reserve a single VLAN for this purpose, permitting multiple VLAN's per subscriber, and even multiple wavelengths; with the least investment in the infrastructure, service providers at this level may also face the greatest competition from companies offering a wide variety of services, some of which may not deliverable with access to only the IP layer.

By combining several of these unbundling options, it is possible that the L2P provider is not the same as the fibre-owner. For example, a specialist middle mile company may rent a wavelength or a VLAN on which to build a L2P reseller service. This is an interesting possibility since it is highly likely that fibre in the first mile will create regional monopolies; this possibility permits competitive middle mile providers to build seamless national networks across multiple regional infrastructures.

It is also conceivable that the operator might permit general IP access to the community network in the subscriber service charge, giving direct access to on-net services such a telemedecine, and community TV. The consumer then chooses an Internet service on the open market. Since the Internet service is likely to be delivered through a PPP protocol, such as PPoE, there can be a clear separation between on-net and off-net traffic, and a proven model for customer authentication.

Issues with PON in open networks

The default choice for many fibre projects is Passive Optical Network (PON) technology, particularly projects led by incumbent operators. While PON is a broad family of solutions, and can generate fierce debate about the pro's and con's of particular family members, there are certain traits which are common to them all which limit the options for creating competitive markets at multiple levels.

The first issue is that PON is designed with contended services in mind; this is a physical trait whereby an optical signal is split and shared among a number of subscribers, typically 32. While this might in some circumstances reduce the installation cost of the network, it permanently removes any chance of offering low contention and uncontended services. It also means that it is only possible to unbundle elements of a PON network after the last splitter. A well known trick of "closed network" operators is to install the last splitter as close the client as possible, even in the basement of buildings, making it economically impossible to build a commercial case for unbundling at the fibre level.

Secondly, each member of the PON family is geared towards a particular next-layer architecture. For example, EPON favours Ethernet. This limits future technology choices, and may leave the future performance of the network solely in the hands of the fibre operator, recreating the same market conditions which led to xDSL being the dictated choice for first generation broadband; any broadband offering today has to work with the circuit-switched network the incumbent specified all those years ago.

Conclusion

While some of these opportunities may not be commercially viable at the start of fibre projects, it is important that contract offerings permit their evolution in the future, and the regulatory regime needs to support this. Today, wave division technology is expensive and can require careful handling but it should be expected that during the lifetime of the fibre, optical technology will dramatically improve. If contracts are offered to providers which give them universal and exclusive access to a subscriber, then it limits future opportunities for competition and may repeat the mistakes of the past, limiting technological innovation in the local loop. Without care, there is a danger of shifting the incumbent problem from the first mile to other parts of service deliver.