NBN Network Design Rules highlight problems ahead

The National Broadband Network design rules are discussed in Business Spectator and the multi-technology mix mess takes centre stage.

Read the full article below

An update of the Network Design Rules (NDR) for the National Broadband Network (NBN) was released by NBN Co on 30 June 2015 to reflect the change to the multi-technology mix (MTM) NBN as required by the government’s April 2014 Statement of Expectations.

Having cleared NBN Co’s renegotiated agreement with Telstra the Australian Competition and Consumer Commission (ACCC) has accepted the amended Special Access Undertaking (SAU) that includes the NDR.

To understand the changes that have been made to the NDR it is important to clearly identify what the Abbott government wants NBN Co to do and what the NBN endgame is. The amended NDR should not come as a shock but now that it has been released it will still be a sledgehammer blow to proponents of the all-fibre fixed access NBN.

What the government wants

The “Fast. Affordable. Sooner.” NBN described in the government’s pre-election NBN plan is underpinned by the principle that government should not have taken a direct role in the provision of broadband and therefore until it is possible for the government to disaggregate and sell off the NBN a lowest cost approach is to be adopted that will ultimately provide little more than an incremental upgrade to the existing broadband access networks.

The major changes to the network design aimed at lowering the cost and speeding up the NBN rollout have been hampered by the very predictable regulatory delays that a major change in direction would introduce and the government’s pre-election promise that under the Coalition’s NBN all premises will have “download speeds of between 25 and 100 megabits per second by the end of 2016 and 50 to 100 megabits per second by 2019” is looking very shaky.

The government has carried out a Cost Benefit Analysis (CBA) of the NBN using technical predictions and assumptions that do not stand up to even the most basic scrutiny. The result has been that the government’s arguments in support of the MTM NBN have relied on a flawed CBA.

Evidence that the CBA is flawed and the MTM NBN Total Cost of Ownership will be significantly more than what we’re being told should have been uncovered by a Life Cycle Cost and Performance Analysis but this key engineering study has not been carried out. A Life Cycle Cost and Performance Analysis is mandatory for large Defence projects so why not for a $43 billion NBN?

The Abbott government was critical of the lack of detailed information about the NBN released by the former government, but since the last election the Abbott government has in many ways actually reduced the amount of information available whilst releasing technically meaningless information such as that provided by the MyBroadband website.

It is now five years after the NBN rollout commenced and the rollout appears to be doomed to go beyond the 2020 expected completion date.

Low cost and completed by 2020

NBN Co is now operating within a strict regime to complete the NBN by 2020 and to adopt the lowest cost rollout approach and this means that technologies, network design and other aspects of the rollout are being modified to meet the key requirements.

And it is important to remember that the current NBN Co management inherited a flawed construction approach, a failure to include HFC in the NBN as a stepping stone to an all-fibre access network during the NBN’s 50-80 year life-time and aspects of the design reflected in earlier versions of the NDR that are a concern. To blame the previous NBN Co management team would be wrong because problems they faced were due to directions from the former government.

To reduce the Total Operating Cost gone from the NDR are the principals of increased reliability and resiliency, improved traffic class management, reduced congestion and lower 50 to 80 year infrastructure life cycle cost that underpinned the previous NBN design.

For some the revised NDR describes a can and string approach that has been adopted to meet the government’s lowest cost expectation. Others argue that the “gold plated” former NBN design is unwarranted. Technical arguments often hinge on small details and the choice between when to overbuild or to upgrade often divides opinion.

The NDR is now in its fourth version and provides an overview of the NBN design to be rolled out to meet NBN Co’s obligations under the regulatory wholesale pricing commitment made in the SAU, which was developed by NBN Co meet two objectives that include:

• giving NBN Co's customers, their end-users and NBN Co certainty about the terms of access to NBN Co's services, including an appropriate regulatory oversight role for the ACCC; and
• providing the framework necessary for long-term cost recovery.

NBN Co states that the “SAU will shape price and non-price terms until 2040, while the Wholesale Broadband Agreement (WBA) is a commercial contract between NBN Co and its customers, and currently has a standard term of 2 years.”

Changes to the SAU, NDR and WBA over the past five years have been substantial and certainty continues to be unattainable as the NBN evolves.

Most significant design change

Probably the most significant change has been the introduction of copper based access network technologies including Fibre to the Node (FTTN), Fibre to the Basement (FTTB) and Hybrid Fibre Coax (HFC). From a technology selection perspective as more technologies are introduced into the NBN the OPEX and life cycle costs increase, irrespective of any savings that may be made during the initial rollout.

But if your goal is to complete the rollout by 2020 utilising a lowest cost approach then the introduction of copper based network technologies is pragmatic, as the increased OPEX and the inevitable need for an immediate technology upgrade or overbuild are not the radar now.

The number of access technologies being used by NBN Co has now increased from three to six and this will increase operational expenditure (OPEX). OPEX costs have also increased due to the government’s direction that NBN Co disaggregate the NBN access technologies which leads to separate technology, management, maintenance and operations silos.

In addition to the increased operations, management, maintenance and information systems costs supporting six access network technologies the use of the copper access network (CAN) is likely to increase OPEX by about $800 million per annum and this cost is expected to increase with the CPI.

The cost benefit versus life cycle cost balance has been tilted towards a minimal initial rollout cost and growing OPEX costs and the need for a technology upgrade or overbuild in 2020 is left for the expected new owners of the disaggregated NBN technologies.

SAU does not provide certainty

The SAU accepted by the ACCC in December 2013 is about 100 pages shorter than the earlier version and the ACCC states "The accepted SAU was submitted by NBN Co on 19 November 2013 and is based on a variation to the SAU submitted by NBN Co in December 2012." Sections that have been deleted include Regulatory Oversight, Service Levels including network performance and availability, Measurement and Corrective Action, Operational Targets and Service Level Commitments.

When a key document has about 25 per cent of the content removed an explanatory note should be provided and the ACCC’s statement of the final decision to accept the December 2013 SAU does not include this guidance.

The SAU, including the NDR, provides limited valuable technical information and appear to be focused on cost, variations, information restrictions and risk minimisation.

An example of the extent that the SAU and NDR are focused on risk minimisation is provided by Clause 1E.6 which identifies situations when NBN Co might need to vary the NDR and provides for the ACCC to be notified of the details up to six months after the variation has occurred.

It can be argued that Clause 1E.6.1 provides flexibility needed by NBN Co to “get on with the job” and not have further regulatory delays to what has already seen three years of the five years to date lost through regulatory review. But we need to remember that the last 18 month regulatory delay was the government’s choice and should not be incorrectly identified as being caused by the ACCC.

SAU clause 1E.6.1 “Scope of Network Design Rules” states:

Subject to clause 1E.6.4, NBN Co will ensure that, on and from the SAU Commencement Date, the Network Design Rules for the Relevant Assets satisfy the following design scope:

(a) in respect of the NBN Co Fibre Network:

(i) the NBN Co Fibre Network is to have a network footprint that is consistent with the coverage obligations set out in the

Statement of Expectations as at 17 December 2010;

(ii) the NBN Co Fibre Network is to be primarily designed and built using GPON architecture;

(iii) the NBN Co Fibre Network is to be capable of delivering the speed requirements specified in the Statement of Expectations; and

(iv) NBN Co is to comply with the Statement of Expectations and the Australian Government’s legislative and policy requirements in respect of the deployment of fibre in greenfield locations, including any alternative models contemplated for such deployments;

(b) in respect of the NBN Co Wireless Network and the NBN Co Satellite Network:

(i) the NBN Co Wireless Network and the NBN Co Satellite Network are to have a total network footprint that is consistent with the coverage obligations set out in the Statement of Expectations as at 17 December 2010;

(ii) the NBN Co Wireless Network is to be capable of delivering the speed requirements specified in the Statement of Expectations and by the Australian Government;

(iii) the NBN Co Satellite Network is to be capable of delivering the speed requirements specified in the Statement of Expectations and by the Australian Government, including an interim

satellite solution with a 6 Mbps PIR downlink Data Transfer Rate; and

(c) in respect of the NBN Co Networks and the Relevant Assets generally:

(i) the Relevant Assets are to have a network availability that meets any applicable law;

(ii) the initial location of POIs within the NBN Co Network will be those identified in the POI List as at the SAU Commencement Date;

(iii) NBN Co is to use existing infrastructure for the NBN Co Network where it is economically and technically feasible to do so;

(iv) there is to be a path for technology upgrade of the Relevant Assets, to the extent possible; and

(v) other matters set out in the Statement of Expectations are to be addressed to the extent applicable

Clause 1E.6.2 “Permitted Variations from Network Design Rules” states:

(a) NBN Co may vary, change, augment or enhance the design, engineering or construction of the Relevant Assets from that specified in the Network Design Rules where such variation, change, augmentation or enhancement:

(i) is contemplated by, or made pursuant to, the Network Design Rules; or

(ii) improves the performance or functionality of the Relevant Assets and results in the same or lower Total Cost of Ownership; or

(iii) achieves savings in the Total Cost of Ownership; or

(iv) is reasonably necessary to establish and maintain the quality, reliability and security of the Relevant Assets or the supply of the Product Components; or

(v) is required in connection with a Force Majeure Event; or

(vi) is required in order to comply with the Statement of Expectations, or a legal, policy, regulatory or administrative requirement, or any requirement of the Shareholder Ministers; or

(vii) relates to the maintenance, replacement or re-routing of assets that comprise the NBN Co Network that has a substantial primary purpose other than the augmentation or extension to such network (e.g. straight swap out of assets for assets as part of routine maintenance); or

(viii) subject to clause 1E.6.3(a), is the subject of an assessment by NBN Co (made at the time NBN Co becomes aware of the need for such variation, change, augmentation or enhancement) that the estimated Capital Expenditure incurred in connection with the relevant variation, change, augmentation or enhancement is likely to be less than the Minor Expenditure Limit; or

(ix) is required to address an urgent and unforeseen network issue where it is necessary that the variation, change, augmentation or enhancement is operational within 6 months of NBN Co becoming aware of urgent and unforeseen network issue and:

(A) the event or circumstance causing the required variation, change, augmentation or enhancement was not reasonably foreseeable by, and was beyond the reasonable control of, NBN Co; and

(B) a failure to implement the variation, change, augmentation or enhancement is likely to materially adversely affect the safe and reliable operation of the NBN Co Network or the supply of the Product Components, Product Features, Ancillary Services or the Facilities Access Service,

(each a Permitted Variation).

(b) NBN Co must ensure that each Permitted Variation is designed, engineered and constructed with the objective of achieving the lowest Total Cost of Ownership.

How can NBN Co argue for a change to the NDR under Clause 1E.6.2 (a)(iii) when the actual total cost of ownership is unknown for a ten, twenty or thirty year period and what does the statement “reasonably necessary” in Clause 1E.6.2 (a)(iv) actually mean? The legalistic gobbledygook and hedging when performance, quality, reliability and security might impact on the bottom line do little to install confidence in the SAU and NDR from a technical perspective.

Significant network design changes

The shift away from an all-fibre access network reduces reliability, if for no other reason than the existing copper will be used and much of this copper is beyond its intended life-time. The increased resiliency that was afforded by a loop Distribution Fibre Network (DFN) rollout has been lost because the updated NDR introduces a star DFN and states that “star DFN deployments will be the default design used for all fixed access networks, providing a more cost effective solution.”

HFC and Digital Subscriber Line (DSL) technologies have traditionally been rolled out using a star topology for cost and time to deploy reasons but as the Warrnambool Exchange fire clearly demonstrated is the need today to actively introduce reliability and resiliency into all sections of the digital network and as a result of the exchange fire Telstra have taken steps to introduce increased loop topology resilience into its network but key facilities remain as single points of failure affecting the NBN.

Australians are well aware of the limitations of the CAN and star topologies and the significant downtime because of the lack of reliability and resiliency was a key reason why the original NBN design included an increase in the use of loop topology throughout the NBN.

In recent times an emerging focus on reliability and resiliency has led to increased use of loop topologies for network design in the backhaul, transit link and distribution segments of the network. NBN Co’s decision to utilise a star topology reduces cost and deployment time but also reduces reliability, resiliency and ultimately network performance.

The introduction of key Layer 2 capabilities have been sidelined beyond a focus on provision of best effort traffic and a committed information rate focused quality of service for voice traffic. The ability to increase reliability by utilising a common high quality network terminating device at customer premises has also been lost with the cost saving move for FTTN and FTTB customers to provide their own NTD.

Brevity and missing information

The brevity of the NDR makes it difficult to clearly see many aspects of the technical design surrounding the FTTN, FTTB and HFC networks and this is arguably because there are aspects of the introduced technologies that are only now becoming available, and this particularly applies to HFC and point to point fibre connections.

From what is provided it is difficult to gain a clear understanding of the HFC network design including how many premises will be connected to a cable run, how much of the current fibre in the HFC network will be leased from Telstra and Optus, and what NDR modifications will be necessary to accommodate DOCSIS 3.1, which has been announced by the Minister for Communications Malcolm Turnbull and NBN Co for implementation commencing in 2017.

The HFC design presented in the NDR is unacceptable for many reasons and the failure to provide any guidance as to the design for utilisation and capacity highlights the brevity and missing information in the NDR. Similar concerns exist for the point to point fibre connections.

Capacity constraints

A simplified view of the FTTN fibre access node (FAN) design is shown in Figure 20 with four fibres allocated for the link from nodes to the FAN and two fibres used initially to provide two 1 Gbps Ethernet connections. As load increases the extra two fibres can be used to provide four 1 Gbps Ethernet connections and if necessary the interface cards can be replaced with 10 Gbps Ethernet cards.

The FTTN nodes might be anywhere between 50 to 100 per cent utilised as the FTTN nodes are likely to be placed near the old telephone cable pillars which typically included up to 200 copper connections for premises within about one kilometre and the FTTN nodes can service about 384 connections.

What this means is the FTTN nodes might not fully utilise the 384 ports available but in the extreme case where 384 premises were connected to one node and customers were accessing the Internet using a number of devices simultaneously, which is often the case between 5pm and 9pm then congestion is likely to occur.

There is no capacity wiggle room in the FTTN design presented in the NDR and the entry of Netflix into the Australian market along with new online education, health and social media services will place an increasing load on the FTTN to the point where there reasonable argument for NBN Co not to rollout FTTN or to immediately shift to a design that is more suited to higher capacity demand.

A similar argument could be made about the microwave transit links in the fixed wireless networks. Up to sixty premises could be connected to a fixed wireless tower sector and there are three sectors per tower. The NDR states that “the three sectors of an eNodeB (maximum of 180 End Users) typically provide an aggregated CIR throughout of approximately 180 Mbps. This determines the Microwave Transport backhaul capacity required.”

“A Microwave Hub Site may have an eNodeB present. The Microwave Transport equipment at a hub site can communicate with various microwave sites and provides an aggregate point for connectivity to a FAN site. The maximum bandwidth planned is 900Mbps, allowing for the aggregation of up to 8 eNodeBs.”

What this means is that the microwave distribution for the fixed wireless is based around the requirement to ensure that there is 1 Mbps CIR for each customer for TC-1 telephony services. The NDR goes on to state that the Wireless Packet Data Network Gateway supports up to 50,000 premises over four 10 Gbps Ethernet connections which again is approximately 1 Mbps CIR.

The fixed wireless network as presented in the NDR is likely to become congested very quickly as customers are added to towers. The fixed wireless design has not changed significantly in the new NDR and highlights the real possibility that capacity will become a significant problem during peak hours.

Congestion in the distribution links appears to be a significant weakness of the NBN that shines like a beacon in the NDR and the FTTN, HFC, Fixed wireless and satellites are all likely to be similarly under-resourced. It could be argued that the data usage growth experienced over the past couple of years will plateau but there is no reasonable evidence of such an event occurring.

Haven’t we learnt anything from the ADSL2+ and HFC night-time crawl that Australians have suffered for many years now when during the peak hours of 5pm to 9pm the network congestion reduces performance significantly?

The significant cost and time required for NBN Co to upgrade capacity as customer satisfaction plummets calls into question the network design described in the NDR and the introduction of FTTN when similar problems with a lack of capacity and significant congestion has been an ongoing problem in the UK.

Further clarification needed

The vast gulf between the previous and current NBN designs, the capacity problem and the prospect of future changes to the NBN as the reality sets in that the current design is deficient, will not meet end-user expectations and will cost significantly more over a 30-50 year period will leave NBN Co’s customers with a great deal of uncertainty and additional cost.

The revised NDR provides some answers to what we should expect by 2020 but the brevity of the NDR and the missing information leaves many questions unanswered so we should expect yet another revised NDR next year.