NATIONAL BROADBAND NETWORK Wireless and Satellite Access Introduction The Government’s National Broadband Network (NBN) announcement on 7th April 2009 stated that, in addition to the 90% of Australian homes served by fibre, the remaining 10% would be served by other “wireless” distribution techniques. It can be expected that many of these homes to be serviced by “wireless” will be in rural and remote areas, currently disadvantaged for communications services in general. Getting the mix of NBN services right for these customers is of vital importance, to ensure the economic and social development of rural and remote areas so that Australia as a whole can leverage the benefits expected from this huge national investment. This paper will look at the key factors influencing the success of deploying wireless broadband, and identify some of the areas where policy decisions will need to be made – some of them being urgent. Note: throughout the rest this document, we’ll use the term “wireless” to indicate terrestrial bidirectional technologies such as 3G, WiMAX and LTE, as opposed to “satellite” connections, even though both technologies are wireless in nature. We’ll refer to the two groups together as “non-fibre”.

Spectrum and the “Digital Dividend” For any wireless access strategy, spectrum planning will be of key importance. Availability of appropriate spectrum determines range, maximum user density, access speed, and how performance falls off as distance from the transmission tower increases. The upcoming availability of spectrum at 700 MHz has sparked a major debate, with mobile operators, terrestrial broadcasters and the NBN itself all candidates to utilise this asset. This spectrum, which will be made available around 2013 due to analogue TV transmissions being discontinued, would definitely suit an NBN wireless solution. It has the potential for high data capacity and also provides a good compromise for range as shown in the following diagram.

WiMAX 3.5 GHz

GSM 1800 MHz UMTS 2.1 GHz WiMAX 2.3 GHz potential LTE 2.6 GHz

Figure 1: relative wireless coverage for different frequencies

GSM 900 MHz UMTS 900 MHz W-CDMA 850 MHz potential LTE 700 MHz

Land Mobile 450 MHz potential CDMA/LTE 450 MHz

700 MHz has propagation qualities that are particularly well-suited to rural areas and also provide good in-building coverage. Conversely, mobile operators could also use 700 MHz for non-NBN services while terrestrial broadcasters are reluctant to give up their spectrum assets, especially if they will incur costs associated with TV channel re-assignment. But as the move away from analogue TV broadcasting eliminates the need for analogue UHF translators across Australia (currently widely spread throughout the 700 MHz band), a large block of this spectrum could be re-allocated into contiguous wide channels (up to 20 MHz), suitable for broadband access. One of the key enablers for this is the ability for multiple neighbouring digital TV transmitters, for example repeaters, to broadcast on the same channel without interference. This principle, called “single frequency network” has been part of the design for digital TV from the outset, with the specific aim of avoiding the wasteful use of spectrum in areas where a single TV transmitter cannot serve the whole community. Spectrum management issues necessarily take a long time to be decided, and if existing services are impacted, a long time to implement. Rolling out the wireless part of NBN quickly might require compromises in terms of optimum spectrum, and associated technology. Planning NBN wireless as part of the wider strategy for spectrum in Australia is perhaps one of the most urgent tasks facing the government and its regulatory bodies. Adequate long-term planning and leveraging global frequency harmonisation is essential to avoid fragmenting the sale and use of the spectrum.

Wireless access speed and capacity claims The government has stated that wireless coverage will be “12 Mbit/s, bursting to 100 Mbit/s or more”, implying a uniform minimum level of service, regardless of the user’s location. However, in a practical deployment, wireless network access speeds can vary greatly according to the distance from the base station – very similar to the limitations seen with DSL today. There are also many other factors influencing wireless access speed and capacity, including terrain, clutter, density of active users and traffic type (like many other technologies, wireless is a shared medium).

One common wireless practice to help achieve high data rates for individual users is to minimise the signal interference through the use of high gain directional antenna systems. In rural areas, this would be the norm for customer premises equipment installation. Furthermore, effective coverage extension in rural areas (>15km from the transmitter) necessitates increased antenna heights to compensate for propagation losses due to the earth’s curvature. Such practices are typical of the robust network engineering that needs to be considered for rural wireless deployments with low user density. Supporting a small number of users, at long distances from the transmitter, results in quite different total cost per user compared to “normal” situations where a transmitter is placed close to a community and backhaul is the only major concern. It may make extended use of fibre more attractive, or drive a mixed strategy of wireless and satellite technologies.

Site 1

Site 2

Legend Throughput High

Low

Figure 2: varying data speeds achievable for a mobile device relative to base station (transmitter) location

Evolving wireless access technologies In order to provide the most consistent service at a reasonable cost, it’s highly desirable that NBN Co. chooses a single technology (although not necessarily vendor) to implement its wireless solutions. LTE is often discussed in the context of a “final solution” for wireless, but in fact, LTE is just one option that continues an evolution towards a common global standard. The following chart shows how existing wireless technologies compare in terms of latency and maximum throughput (downlink peak speed only), two of the most important characteristics of a wireless broadband system. It also shows where these technologies will evolve to in the near term, based on current standardisation activities. It’s easy to see that all three of the wireless technology standards would be able to operate in the ball park that NBN requires. However, each of these technologies also has a longer term evolution to a new common standard known as 4G (IMT-Advanced), which promises performance that aligns closer to NBN’s stated objectives. 100 ms

65 ms

60 ms

Latency

50 ms

50 ms

14 ms

EV-DO EV-DO RevA RevB 1.25MHz 5MHz

UMB 20MHz MIMO 4x4

CDMA

10 ms

WiMAX WiMAX 16e 16m 10MHz 20MHz TDD MIMO 4x4

10 ms

HSDPA 5MHz

WiMAX

HSPA+ LTE 5MHz 20MHz MIMO MIMO 2x2 4x4

3G

300 Mbit/s

WiMAX WiMAX 16e 16m 10MHz 20MHz TDD MIMO 4x4

14 Mbit/s

HSDPA 5MHz

42 Mbit/s

HSPA+ LTE 5MHz 20MHz MIMO MIMO 2x2 4x4

IMT-Advanced 100MHz MIMO 4x4

QoS implementation on wireless is mandatory

Ensuring high performance Although these charts show two of the key attributes for wireless technologies, they are not the only factors that need to be considered. Other characteristics such as range, capacity, Quality of Service (QoS) support, immunity to noise and spectral efficiency (how much data you can get from a certain amount of bandwidth) are all important. Luckily (perhaps), all three systems are roughly aligned in these respects. The bottom line therefore, is that NBN Co. does have flexibility in its choice of wireless technology and can consider “softer” factors such as availability of suitable vendors, likely take-up in other parts of the world, CPE cost, etc. Today, LTE and WiMAX have so far emerged as the prevailing wireless choices in considering such factors. Of course, it would be quite naïve to study charts like Figure 3 and assume that all users could experience the data rates mentioned there. Not only are these peak rates for a shared total bandwidth, with all users in the “sweet spot” for reception, but these next generation technologies routinely quote Multiple Input Multiple Output (MIMO) techniques to achieve claimed peak data speeds. And, unfortunately, NBN’s wireless solution will be used primarily for regional and rural deployments, where the effectiveness of MIMO (which relies strongly on multi-paths typically found in high areas of building clutter) will be significantly reduced. An effective 4x4 MIMO could effectively reduce to a 1x1 system with only a quarter of the peak throughput. Finally, the access, or radio network, is only part of the story when it comes to wireless performance. Regardless of the technology chosen, it will be vital for NBN Co. to implement a flat all-IP underlying wireless network architecture to ensure that characteristics such as latency meet expectations. This will be especially important to enable the support of real time services such as Voice over Internet Protocol (VoIP), as well as some high throughput applications, depending on how the protocol is implemented. A well-engineered network, based on WiMAX (802.16m), or 3G (LTE) and eventually 4G, with minimal hops and nodes, will typically deliver latencies in the order of 10 milliseconds. If this goal is achieved, then wireless customers can look forward to a quality of service that closely mimics the performance enjoyed by customers connected through direct fibre.

As mentioned before, implementation techniques will be key to ensuring a good user experience for wireless broadband. The overall objective is to ensure maximum service compatibility between customers on fibre and non-fibre access techniques. This will allow application providers (with the likely exception of live TV) to provision services to end-users without the need to know whether they are connected via fibre, wireless, or satellite, helping to ensure that rural and remote customers are not disadvantaged compared to those residing in metro areas. QoS will be a key component of fibre-delivered NBN: it’s essential for the simultaneous delivery of multiple services if interference between those services is to be avoided. In a wireless environment, with less capacity, management of that bandwidth via QoS becomes even more important. Evolved wireless access technologies such as mobile WiMAX, 3G HSPA and LTE inherently support robust QoS implementations to ensure differential treatment of services while minimising congestion over the shared wireless access medium. It’s worth remembering that NBN isn’t just about faster internet, and QoS will be essential to ensure that non-fibre customers enjoy the wider benefits that NBN will bring to the community at large.

Wireless network engineering is a complex subject Basic wireless performance characteristics such as coverage, subscriber density and access speed are determined by the overall network design, availability of appropriate base station sites, and spectrum, in particular frequency and bandwidth. As in any engineering design there are trade offs or compromises. For example, higher frequencies such as 2.6 GHz typically have larger bandwidths available – which can support a larger number of customers at potentially higher access speeds – however these higher frequencies typically result in reduced coverage. A likely outcome of this would be a requirement for more base station sites – and these sites are normally the most costly part of a wireless network deployment. Putting in more base stations might mean that backhaul fibre would need to be extended closer to the community, which in turn would add to the number of communities where fibre can be the total solution, thereby eliminating the wireless “problem” altogether. In order to provide adequate service with minimised costs, NBN Co. will need to undertake very detailed engineering and population surveys before bringing broadband to wireless-connected communities. As point of reference, LTE, one of the candidates for NBN wireless, is designed to operate with a channel bandwidth of between 1.4 MHz to 20 MHz, however if narrow bandwidths are deployed, the data rates seen by customers will fall far short of the government’s objectives. If the target sustained access speed is 12 Mbit/s, then the maximum 20 MHz bandwidth of spectrum is likely to be required – and this is inevitably associated with the higher frequencies and the corresponding challenges on the coverage area. The bottom line is that the performance of any wireless system can only be predicted by means of thorough analysis. The two key questions to be answered from a user’s standpoint are: “Will I get enough signal strength and quality?” and “What data speed can I expect?” The first of these can be covered by performing a radio link budget and propagation analysis, where transmitter output level, receiver sensitivity, terrain obstructions and several other parameters are taken into account. However, in order to predict the actual data speed a user will experience, signal strength alone is insufficient. Capacity modelling for the radio sector is also required, given that all the users in that sector are sharing the available data bandwidth. Only performance claims backed up by this type of rigorous modelling should be viewed as valid for predicting the experience that real users will have in the wireless part of the Australian NBN.

Backhaul for wireless customers Backhaul is the term used to define the highspeed connection that links whole communities into the main network. The government has already embarked on a programme of backhaul enhancements to reduce broadband “black spots” within Australia. For wireless-connected communities, there are two key technologies that could be used for backhaul: fibre and microwave radio. Although microwave radio links can operate in the hundreds of megabits per second range, their capacity will always be much less than fibre, and their upgrade costs are very much higher. Fibre will therefore be the backhaul method of choice, subject to cost and usage analysis. One of the key facts limiting wireless for backhaul is the effect of the earth’s curvature: this effectively places a maximum distance of around 40 kilometres between repeater towers. And as three radio hops is normally considered the maximum (for reliability reasons), this implies that communities isolated by more than 120 kilometres cannot be served with a radio backhaul. Over ideal territory, perhaps surprisingly, the cost of laying fibre is not dramatically different to providing wireless backhaul. This is because radio towers are expensive to erect, and the cost per kilometre of laying fibre has fallen dramatically over recent years. However, over tough terrain, fibre costs can rise by an order of magnitude. The implication is that communities beyond the reach of wireless backhaul, and with very challenging surrounding terrain will, for the foreseeable future, have to relay on satellite for their NBN broadband access.

Wireless and satellite technologies will exist side by side, but wireless penetration will increase Broadband access via satellite is generally considered a last resort for areas where fixed line and wireless distribution techniques are impractical. There are a number of reasons for this, including: • Cost and complexities involved with communicating from the user to the network. • Extended round-trip delays which can have a significant impact on the performance of some applications. • A system capacity of only a few thousand users per transponder spot beam –and one beam covers an area of many thousands of square kilometres. • Realistic user data rates limited to a few Mbits/s toward the user and less than one Mbit/s from the user (the higher the data rate the more costly the antenna and associated electronics will be). • Continued high operating cost of access to satellite transponders and a limited satellite lifetime (typically 10 to 15 years) before replacement is required. Despite the limitations in the use of satellite for broadband services, this does not mean it has no role to play in digital future of Australia. Even assuming that wireless-connected customers could reliably achieve broadband rates of 12 Mbit/s sustained, this would not represent a viable delivery vehicle for general TV services. One or two Standard Definition (SD) channels would be the most optimistic target, and reception might be subject to degradation or even interruption due to atmospheric conditions. For broadcast services, especially TV and High Definition (HD) TV, satellite will therefore nearly always be preferable to wireless broadband. This means that non-fibre communities in areas not served well by terrestrial TV broadcasts (a possibly significant overlap) will nearly always need to use satellite, at least for broadcast services and possibly for broadband access as well. However, for general network services such as internet, the expanding footprint of wireless coverage will gradually make satellite-based broadband less and less attractive. Eventually, only a very small percentage of the population will need to rely on satellite for broadband access.

Non-fibre customers will need professional installations Most non-fibre customers will, by definition, reside in quite remote areas. This has installation implications both for wireless and satellite. In the case of wireless, distance to the base station is the issue; for satellite, footprints are (ironically, perhaps) targeted to cover metropolitan areas meaning that reception in remote areas can be marginal. As noted earlier, wireless customers will normally need an outdoor external antenna, aligned towards the serving base station. This will require a professional installation similar to existing free to air broadcast television antenna installations, and, depending on location may also require additional mast installation. Satellite customers will likely require a ground-based dish, one metre diameter (or more), again requiring professional help to install. Rural NBN customers requiring broadcast services will likely need both antenna and dish to provide a complete solution. The bottom line is that non-fibre access is unlikely to be cheap compared to fibre-connected homes, at least so far as installation is concerned. However, local businesses such as TV installation companies may benefit if they are used for commissioning NBN customers. Care will be needed with pricing to ensure that non-fibre customers are not significantly disadvantaged due to high installation costs. Currently, government subsidies are available for customers who require the use of satellite to access internet, and this may continue under NBN.

Wholesale considerations A good deal of discussion has already taken place concerning appropriate wholesale models for NBN. The main discussion has been around the relative merits of a bitstream model, where access seekers connect using a (nominally) single Ethernet interface to service many customers, versus fibre unbundling, where access seekers are responsible for physically managing and lighting the fibre. Bitstream, which offers a much more graceful way of connecting customers to service providers, is likely to be the method of choice for the fibre NBN. In the non-fibre part of NBN, achieving the equivalent of fibre unbundling would pose some quite difficult spectrum and engineering issues and would almost certainly never be economically viable. However, even with bitstream, there are technical issues to be considered. Most significant of these is defining how an individual subscriber can simultaneously connect to multiple retail service providers. Forcing wireless customers to choose just a single retail service provider would simplify the wireless wholesale connectivity problem, but deny wireless consumers the flexibility to choose a mix of applications from a range of retail service providers. Additionally, some kinds of application such as smartgrids, eHealth and eSecurity may only be feasible if they can be provided via an independent bitstream connection to the consumer. The bottom line is that some detailed architectural work, possibly requiring wireless development to introduce additional open access flexibility, will need to be considered so the NBN Co. can offer harmonised fibre and non-fibre bitstream access alternatives.

Glossary CDMA – Code Division Multiple Access CPE – Customer Premises Equipment DSL – Digital Subscriber Line EV-DO – Evolution-Data Optimised (referring to CDMA) GSM – Global System for Mobile communication GHz – Gigahertz HDTV – High Definition Television HSDPA – High-Speed Downlink Packet Access

IP – Internet Protocol LTE – Long Term Evolution Mbits/s – Megabits per second MHz – Megahertz MIMO – Multiple Input Multiple Output

Engineering the wireless part of NBN will be a challenging task. Before the first base station is installed, important decisions regarding start date, radio technology, spectrum, services alignment, detailed performance objectives and wholesale arrangements need to be taken. Satellite will continue to have a role to play, both in areas where TV reception is poor, and wireless is impractical for cost or engineering reasons. In the end, detailed modelling will be required to determine the correct mix of technologies for each neighbourhood, down to street level. The availability of backhaul, population density and local terrain will all need to be considered if the NBN is to provide the best possible service at a minimum cost. Challenges associated with non-fibre access methods may increase the fibre-connected percentage past the 90% forecasted to date.

QoS – Quality of Service RevB – Revision B of EV-DO SD – Standard Definition TDD – Time Division Duplex UHF – Ultra High Frequency

IMT – International Mobile Telecommunications

UMB – Ultra Mobile Broadband

IMT-Advanced – 4th Generation of cellular wireless or 4G

Wireless and satellite will play a vital role in the provision of NBN services to the 10% of the population outside the economical reach of fibre. While it is important to specify minimum standards for this 10%, care is needed to ensure that expectations are set correctly and that excessive costs – including installation at the customer’s site – are avoided.

NBN – National Broadband Network

HSPA – High Speed Packet Access

IMT-2000 – 3rd Generation of cellular wireless or 3G

Conclusions

UMTS – Universal Mobile Telecommunications System

Alcatel-Lucent Australia contact details:

VoIP – Voice over Internet Protocol

Ric Clark, Chief Technology Officer ph: +61 3 9664 3407

WiMAX – Worldwide Interoperability for Microwave Access

Lisa Poninghaus, External Communications Manager ph: +61 2 8306 5645