Friday, November 06, 2009

Designing the National Broadband Network

The Australian Telecommunications Users Group (ATUG) held a National Broadband Network Reference Model Forum in Sydney this morning (there will be another in Melbourne on 12 November 2009). Based on the morning's discussion, my view is that a simpler internet model be used in place of the Communications Alliance model. This will be technically simpler to implement and will also avoid many difficult regulatory issues with telephone and broadcast services. In essence the NBN will be "an internet", which will be part of "the Internet". The NBN can carry many different services using internet protocols, including services which emulate the plain old telephone service (POTS), cable TV and broadcast TV, without being limited to only providing those services or providers.

Paul Brooks, Lead Consultant, NBN Project of Communications Alliance gave a detailed overview of the Communications Alliance High Level Architecture Options for the NBN in detail. His view is that it is likely that the NBN will use a passive optical network for Fibre to The Home (FTTH). This will provide considerable bandwidth to the home, but it is not clear how far this will extend through the network.

While NBN will own and install the connections to the home, there is no architectural distinction between NBN and non-NBN back end networks. The Communications Alliance model operates at level 2 . It is likely that IPv6 will be used or services such as VoIP at the higher layers of the model. Issues to be resolved include how many points should other providers be able to connect at (options range from 5 to 500).

At the home termination, it is not clear how the consumer will connect. Options range from a socket to which any equipment can be connected to an NBN supplied set top box. One issue which I raised early on the the NBN process was the need for operations in disasters during mains power failure. This seems to have been taken up with discussion of who provides the backup batteries.

Stepping back from the details Paul Brooks pointed out a principle should be customer choice. Each service may be connected to different devices from different suppliers ad networks. The example given was a smart electricity meter provided by the energy company, which the householder has little control of but still has to have working over the NBN.

It occurs to me that in all this some points have been lost:

1. Its the Internet: The primary purpose for the NBN is to provide access to the Internet. It is likely the system be implemented using internet protocols. The simplest way to provide Internet access via an internet network is with internet protocols. Therefore the the NBN should be designed as an internet. Much of the Communications Alliance model discussion seems to be about old fashioned connection based network design which is not needed and not relevant for an internet.
2. Layers aren't real: While there s much discussion of Layer 1 and Layer 2, these are abstractions and so of little use for practical decision making. In the discussion it does not seem to be made clear even which multi-layer model is being discussed (the ISO OS
model has seven layers, whereas IP has only five).

Peter Hitchiner, Australian Computer Society – Telecommunications Society of
Australia, gave a more general overview as to what the NBN should do (similar to the ACS talk I gave to ATUG in Canberra). He pointed out that the nature of the NBN service is not clear, in particular is "layer 2" access the preferred industry approach. A major question is will IPTV services be treated equally (a major policy question for the Federal Government).

At this point the forum moved into a discussion to explore some of these issues. This proved very interesting and useful. On the access issue Paul Brooks mentioned that the home access box might have four ports (presumably Ethernet copper cable ports), plus possibly a telephone and a TV port. It seemed that he was envisioned each port would provide a distinct "service" from a separate "service provider". In the subsequent discussion it became clear that the model the Communications Alliance's proposing is to emulate a point to point service over the NBN, on top of an underlying IP network.

Stephen Wright, from Gibson Quai-AAS - Telecommunications Consultants (GQAAS) then talked about the network resilience required. Telstra provides about 99.90% reliability for telephony services (PSTN). Stephen suggested we should aim for 99.95% or higher for the NBN. My view is that it should be relatively simple and inexpensive to achieve this level of service for the NBN for telephony services. This assumes that the NBN is configured to provide different levels of reliability for different services, with emergency services having priority. As an example, the household is likely to want enough bandwidth to call an ambulance in an emergency, and will accept that this should take priority over being able to watch TV. With the NBN configured to provide enough bandwidth to provide TV most of the time, there should be enough capacity to handle lower bandwidth services, such as telephony almost all the time.

The Communications Alliance proposed model is not a good one, is not in the public interest and should not be adopted. Its complexity comes from trying to reproduce the restrictions of the old telecommunications system in order to support old business models. Instead I suggest accepting that the NBN will provide an Internet service. The model then becomes very much simpler, with an unlimited number of service providers able to provide services to the home over one Internet connection. Where a service provider needs a high level of security for their service, for example a telephone, smart meter or a TV set top box, they would need to ensure that the software or hardware they provide to the home has the needed security built in. There is then no need to worry about how many ports to provide or what types. Only one port is needed which can support all the services required.

This would be similar to the electrical sockets provided in the home. These are all worried in parallel and provide the same service. The householder can purchase their own devices to plug in. The householder can also purchase multiple adaptors to plug in. If the householder wants to plug in a refrigerator from a particular supplier, they do not need a special power point installed which only provides one brand of electricity.

The NBN will likely replace the current telephone, broadcast TV and cable TV services. I suggest replacements for these services be done in a way which does not limit the availability of new services. This would be a change in the previous government practice which has been to protect encumbent providers from competition from new services. As an example, the conversion to digital TV in Australia was designed so that the existing analog TV stations retained their oligopoly, even though the new digital technology di not require this.

There is a case for providing telephony and TV as part of the basic NBN service, but architecturally these can be simply services on top of the Internet. In this way the household will not be locked into a new monopoly unnecessarily. As an example, the householder would be able to pay a service provider to provide an ordinary telephone POTS type service from an Australian telecommunications provider. But the householder should be able to use the same NBN equipment to make free Internet calls and to sign up with several other telephone providers, including companies located anywhere in the world, if they wish.

Similarly, the house holder could use the NBN to watch Australian free to air TV and cable TV. However, the householder should be able to also access any other TV-like service available over the Internet from anywhere in the world. The fact that Foxtel might wish to provide a restricted pay TV service using NBN, should not stop other provision of TV like services for free by others.

The issues of telephony on the NBN are not technically complex, compared to the regulatory and public policy issues. TV type services over the NBN are slightly more complex technically, but are dwarfed by the complexity of the public policy issues. There is not sufficient time to work through all of the sectional interests involved in time to implement the NBN. With the current timetable, Australian will still not have transitioned completely from analogue to digital before it will be time to start replacing the digital TV broadcast service with the NBN.

I suggest the Australian Government take the opportunity to short circuit the process by setting some simple goals for the NBN in delivering services over an internet and then let the NBN company get on with the implementation. The NBN is not building a telephone network, nor a pay TV network, it is building a network which can be used for carrying such services.

No comments: