Thinking beyond "copper" versus "fibre"

The triumphant return of Kevin Rudd is unlikely to save Labor from defeat at the federal election. However, he will presumably inject fresh impetus into taking Coalition’s policies to task, and it’s only a matter of time before the National Broadband Network comes into the frame.

While we are still waiting to see who will replace Stephen Conroy as communications minister, there is a real need for Conroy’s successor to dig deeper and move the conversation beyond the 'copper versus fibre' debate.

The NBN debate has so far been firmly focused on the access network which is known as the 'last mile' connection between the customer premise and the local exchange.

But think about it, does it really matter if your home is connected over the access network at 25, 50, 100 or 1000 Mbps? Do we really need fibre, or will the existing copper do?

NBN Co is currently rolling out FTTP (to 93 per cent of Australians) under Labor’s NBN plan. The Coalition’s NBN plan will reduce the FTTP (22 per cent) rollout and utilise FTTN (71 per cent) for the bulk of Australians.

To understand which plan will provide the best outcome for Australia over the next 30 years, we need to identify key criteria against which the Labor and Coalition plans can be measured. To identify the key criteria it's important to consider the NBN architecture as a whole and whether or not it will provide Australia with a next generation network.

The Australian digital network, which includes the NBN, consists of five network segments:

1. International – links and transit points that connect the Australian digital network to the rest of the world. Retail service providers connect to international networks over international links.

2. Backhaul – links and transit points that connect the 121 NBN Points of Interconnect (PoI) that are part of retail service provider networks. Customer traffic to and from retail service providers travels over backhaul from the 121 NBN PoI.

3. Aggregation – The NBN aggregation networks consist of  121 sites containing PoIs and systems that are used to manage traffic as it flows between PoI and the Fibre Access Nodes (FAN). A single PoI connects to many FANs.

4. Transit – The NBN transit network consists of the aggregation network, FANs and the fibre links between them, including DWDM optical fibre systems in regional areas.

5. Access – The NBN fibre access network connects customer premises to the FAN.

NBN Co is responsible for 121 separate networks that start at the PoI and end at the customer premise. This is an interesting situation because the 121 networks operate in isolation.

The key criterion for a next generation national network available in 2014 includes:

1. Minimum access network speeds of 1 Gbps download and 400 Mbps upload

2. Average 100 GB per month per customer increasing tenfold over the next decade

3. Network utilisation below 70%

4. Implement Quality of Service (QoS) utilising four traffic management classes

5. Three backhaul provider connections at each PoI

Connection speeds not enough

Connection speed itself is not sufficient to improve the operation of the digital network. Currently most of the traffic flowing around the Australian network is best effort, which means that the network provider does not guarantee that the traffic will arrive nor when it will arrive – no QoS at all.

This is like running a train network without a timetable – congestion will quickly occur and the network will eventually come to a halt. The Australian digital network follows this scenario daily, particularly in the peak period between 5-9pm. Most home ADSL users find network speeds, connection reliability and performance to be quite poor in the early evening when the family sits down to catch up with online activities.

To improve traffic flow and QoS across the network there is a need to introduce traffic classes and traffic class management. The idea for traffic class management has been around for many decades yet carriers have been reluctant to implement it other than for business customers who will pay a premium for an improved Service-Level Agreement (SLA).

The cost of implementing traffic class management today is negligible, so it is time that this occur.

NBN Co, working with the service provider industry, has identified four traffic classes which include:

1. TC-1. Examples of traffic that fall into this category include voice, control, emergency services

2. TC-2 and TC-MC (MC means multicasting). Examples of traffic that fall into this category include streaming standard and high definition video

3. TC-3. Examples of traffic that fall into this category include premium data, gaming, business Virtual Private Network (VPN) access

4. TC-4. Examples of traffic that fall into this category include all general internet access

NBN Co has already implemented TC-4, which is the equivalent of best-effort traffic management, and TC-1, which is being used for internet telephony, control and emergency services. TC-2 and TC-3 have been identified as traffic management classes that would benefit business and have been slated to be implemented between 2013 to 2015.

It would be wrong for TC-2 and TC-3 to be offered only to business customers. Non-business customers would benefit from improved traffic class management, particularly for real-time, multi-user applications like games, video conferencing, social media and others.

The NBN Co traffic class management approach builds upon the need for improved QoS not only for business customers but for all customers. When NBN Co has fully implemented the four traffic classes across the NBN there will be pressure on backhaul providers and retail service providers to implement the same traffic classes within their networks.

It’s fair to assume that there will be industry resistance to generally implementing traffic class management as the provision of premium SLA has been a cash cow for the telco industry for the last decade.

Another way to see how traffic class management works is to consider how vehicle traffic is managed on urban highways today. Highways provide a limited vehicle carrying capacity and during peak periods highways often turn into car parks – a bit like the digital network between 5-9pm. Authorities came up with the idea of setting aside lanes for buses, taxis and vehicles with more than one occupant to use during peak periods. This traffic management approach means that the highway becomes more efficient especially in times of high utilisation.

Maximising potential

Until the four traffic classes are fully implemented the NBN will not be able to reach its full potential, satisfy this key criterion for a next generation network, and thereby move beyond the best effort networks found today.

Networks must have sufficient capacity to service traffic demand. There are several approaches to work out how much capacity the network needs, but invariably the provision of network capacity is a major cost and the percentage total capacity versus offered traffic falls over time. What this means is that the network becomes increasingly congested over time.

NBN Co’s philosophy is to avoid being the bottleneck in the end-user’s experience, and allowing the access seeker to determine their end-user experience through mechanisms within their own control - dimensioning within their own networks, and dimensioning of NBN Co services.

The NBN Co approach (for its fibre peak information rate services) is to use a capacity threshold of 70 per cent as a trigger for an upgrade to add more capacity. NBN Co will monitor its transit network and reassess shared network capacity allocation if the average utilisation exceeds 70 per cent of Layer 2 capacity for 30 minutes at least three times within any consecutive seven day period.

This means that extra capacity would be made available to a network segment if the capacity threshold trigger occurs. In some circumstances there would be a delay before the extra capacity is added due to the need to add or upgrade equipment or transmission capacity, but this action would be traceable using NBN Co’s capacity threshold trigger approach.

A key aspect of the retail service provider backhaul network is the need for adequate capacity and competition otherwise this part of the network will become a costly bottleneck.

There must be a minimum of three backhaul providers for every PoI and backhaul capacity should have a capacity threshold trigger of 60 per cent of offered traffic at a PoI. Currently the Australian Competition and Consumer Commission backhaul criteria is for two backhaul provider connections at every PoI, but this is insufficient for effective backhaul competition.

Tasmania was an internet blackspot in the days when Telstra provided the only backhaul link across Bass Strait. More recently with the introduction of the Basslink backhaul connection the situation for Tasmanians is no longer dire but it is still unacceptable.

The ACCC stepped in and determined that backhaul charges to Tasmania are not to exceed 40 per cent above that for the same distance between regional centres on the mainland. While the ACCC’s intervention was necessary it has effectively fixed the price and did not take into account congestion and minimum capacity thresholds.

Giving the public what they want 

Australians have discovered online media over the past couple of years and we have witnessed explosive growth in online video repositories like the ABC’s iView. Data usage growth is growing rapidly and there has been an unfortunate side-effect which is the daily question, “How much data do I have left this month?”

Bill shock is a nasty event that is only being reduced as the telecommunications industry is forced, kicking and screaming, to clean up its act. For this reason the telecommunications industry is unloved and seen to be a blight on society but a necessary evil. It is time to put bill shock into the past and that requires an average data allowance that reflects society’s needs.

An average 100 GB per month is not unreasonable for an internet user today and it should be expected that this figure will increase tenfold within 10 years. There must be adequate capacity within Australia’s digital network to handle ever increasing data requirements.

If NBN Co implements the four traffic classes by 2015 then backhaul and retail service provider network capacity will become an issue – because adequate capacity is an underlying requirement for QoS and traffic class management.

Finally we come to the issue of the end-user connection speed. Without adequate network capacity, QoS and traffic class management there will be little difference between the NBN’s fibre access network and the existing copper based access network.

Ultimately, Australia’s digital network will not improve unless the entire telecommunications industry embraces change and jointly facilitate the implementation of a next generation network. Improving the network is the promise from both Labor and the Coalition and the viable implementation of traffic class management needs to be a considered.

Both sides of the debate have spent far too much time spinning around in circles on the copper versus fibre issue and if Malcolm Turnbull and Kevin Rudd are really serious about building the NBN to expand our innovation horizons then it’s time to start looking at the bigger picture.

Mark Gregory is a senior lecturer in Electrical and Computer Engineering at RMIT University.