The truth about NBN Co’s satellite needs

56

analysis Does the National Broadband Network Company really need to launch two expensive new satellites to provide remote Australia with broadband? Setting the politics aside, from a technical perspective, it appears the answer is a clear: “Yes”.

This week saw NBN Co announce it had selected satellite specialist Space Systems/Loral to build and launch two new satellites that will provide high-speed broadband to Australians. The $620 million contract will see the satellites target the three percent of Australians who are not slated to receive either fibre or wireless broadband under the NBN policy, and will come into play in 2015 when the satellites launch.

Although Prime Minister Julia Gillard and Communications Minister Stephen Conroy used a nationally broadcast press conference to emphasise that the deal was necessary, and that NBN Co had gone through an exhaustive two-year satellite investigation and procurement process, it was immediately attacked by Shadow Communications Minister Malcolm Turnbull, who sees the deal as being too expensive. Turnbull’s core claim is that NBN Co could rent satellite capacity, instead of building its own. “There is enough capacity on private satellites already in orbit or scheduled for launch for the NBN to deliver broadband to the 200,000 or so premises in remote Australia without building its own,” Turnbull said in a statement.

However, the truth of this matter, as far as Delimiter has been able to ascertain, is that Turnbull is incorrect on this matter — as the Coalition has unfortunately been several times over the past month with relation to other NBN matters, such as the issue of whether cutting the NBN would save the Government money, or whether broadband prices are slated to rise under the NBN.

To ascertain the technical truth of the satellite situation, Delimiter spoke with NBN Co’s project director Matt Dawson on Friday afternoon.

Dawson pointed out that NBN Co had already followed Turnbull’s advice and leased a substantial amount of space on existing commercial satellites owned by IPstar and Optus. That deal, signed in May 2011, already saw a remarkable improvement in the satellite services available to Australians living in remote areas. It has delivered speeds of up to 6Mbps to residents and businesses which previously could attain much less — leading to “terrific” customer feedback, according to Dawson. However, the executive pointed out that just launching that interim solution had already soaked up “the lion’s share” of the commercially available satellite capacity in Australia. “Even that hasn’t got anywhere near the capacity we need in the long term,” he said.

To understand why, it’s necessary to both look at the technical details of satellite capacity, as well as the practical outcome of the interim satellite service launch.

For starters, due to its capacity limitations, the interim service is only capable of covering some 48,000 Australians, while there are several hundred thousand residents which sit within that three percent total not slated to receive fibre or wireless broadband. And secondly, where it does reach those people, the capacity constraints means the service, while reaching decent 6Mbps speeds, only features miniscule download quotas — currently averaging around 6GB per month, according to Dawson.

If you look at the NBN satellite plans offered by companies like HarbourSat and Active8Me, you can see this trend in action. HarbourSat’s top plan offers a tiny 10GB of peak and off-peak data, while Activ8Me offers higher plans, ranging up to 31GB — although at prices like $99.95 per month. For those prices, Australians in the fibre NBN areas will be able to download something like a terabyte of monthly quota.

Dawson points out there are “huge gaps” between the broadband service which rural Australians can receive through satellite, and the service available in city areas. “Doesn’t take long to use up that sort of capacity,” he says of the 6GB quota plans. “So the argument needs to move away from peak speeds, to what’s the real capacity of the network — what are people really getting.”

In terms of real capacity available on the market, Dawson says that NBN Co could probably kluge together something like 6 to 10GHz of capacity through commercially available agreements. The total capacity available to Australia was something like 40GHz, he said. Part of the problem is that the current satellites over Australia also service other countries in the Asia-Pacific region. In comparison, NBN Co’s brand new satellites will be able to deliver something like 90GHz, as they will deliver dedicated capacity to Australia.

In addition, much of the commercial satellite capacity in Australia is targeted over metropolitan areas, where the greatest return will be made for commercial operators, due to a heavier concentration of targets. In comparison, NBN Co will explicitly target users in remote areas, where normal satellite operators would find it hard to operate a commercially successful service.

The impact this rollout will have on the nation’s rural broadband problem is dramatic. For starters, speeds will be boosted immediately, from the currently available 6MBps (or even less in many areas) to 12Mbps. Coverage will also be boosted — from something like 48,000 people to several hundred thousand, including Australia’s external territories such as Norfolk Island, Christmas Island, Macquarie Island and the Cocos Islands. And the monthly quota problem will also be resolved, due to the excess capacity available through the infrastructure.

There’s also another question to be asked: Why launch two satellites? Couldn’t one do? No, according to Dawson. NBN Co is launching two satellites for redundancy, and to spread the load.
Building satellite broadband infrastructure isn’t like building terrestrial broadband infrastructure. If something goes wrong in space, NBN Co won’t be able to visit its satellite installations to fix the problem. “Anyone designing any telecommunications networks, must design in diversity into the solution,” Dawson says, noting satellite launches are not without risk. In the case of one satellite partially or totally failing, the several hundred Australians being served by the infrastructure won’t lose signal.

Instead, he says, NBN Co will be able to migrate all the users off one satellite and onto the other. Under normal circumstances both will be used, with each taking some of the load of Australia’s telecommunications needs.

At the end of the day, whether you believe Dawson and the crew at NBN Co is dependent upon your knowledge of the satellite industry and your ability to trust NBN Co’s engineers. However, speaking with the executive, it is clear that Dawson himself has just completed an exhaustive process of several years’ effort investigating Australia’s current satellite capacity and how NBN Co should best meet its government policy demands. “I know you want simple answers, but the devil is always in the detail,” he says at the outset of our conversation. “We’ve taken two years to go through this every which way, [seeing] what’s available, what’s becoming available, where the bandwidth is laid down. Is it C-band, is it I-band, is it KA-band.”

The result of that investigation, Dawson said, led NBN Co increasingly to the answer of launching its own satellites. “The more and more you go into it, the more and more systems engineering you need to go into to use what little capacity there is anyway,” he says. “We have gone through this upside down and backwards. We know where there’s capacity.”

Solving the problem through leasing capacity, Dawson said, “just didn’t add up”. “The maths is pretty straightforward,” he added. “Network architects, network engineers do the maths [and] they know how to design networks. That’s what we’ve been doing for the past two years. When you start to examine what capacity they’re talking about, the answer starts to become obvious.”

NBN Co has also been criticised for its procurement process for the satellite process. However, Dawson noted that the two process had been “very competitive”. “We know we’ve got very good prices for these sort of satellites,” he said. “We really do try to do the best that we possibly can. We have to be fiscally responsible, with taxpayers’ dollars,” he adds.

Debating satellite telecommunications costs and prices is not an easy business. There are relatively few experts on the issue in Australia. This week Delimiter called several noted telecommunications specialists, but found it hard to get objective advice on whether NBN Co had followed the correct path in deploying its own satellites.

However, I believe Dawson when he says NBN Co’s investigation of the satellite industry led it irrevocably down the path of launching its own infrastructure. The executive spoke passionately and at length about the matter, and had clearly been enmeshed in the intricate details for several years.

Like many of those who work at NBN Co, Dawson’s LinkedIn profile reveals that he’s a seasoned and respected technology executive who’s spent time working with both Australia’s private and public sector over the past twenty five years. From 1993 through 1997, Dawson was a senior project manager at Telstra. At that time he was working on large scale communications programs for the Department of Defence. Later, the executive ran his own consulting business, which was purchased by local IT services group SMS Management and Technology.

After that point, Dawson became Tabcop’s general manager of group infrastructure and operations, before joining NBN Co in April 2010, in what was at that time one of the company’s earliest intakes of employees. He was at that time a program manager for NBN Co’s networks area, but was quickly promoted to project director of the satellite division in July 2010.

Now, I’m sure it’s possible to get other views of this situation. Turnbull, for one, has stated that he has spoken directly to the satellite industry about the issue, and is convinced that NBN Co can lease the capacity it needs to provide, instead of building its own satellites. In addition, it is possible to argue that from a policy perspective, remote Australians don’t need 12Mbps broadband speeds or higher quotas, and could get by on lesser broadband solutions, given that they’re living in such remote areas.

However, for what it’s worth, I believe Dawson when he says NBN Co couldn’t meet its current policy commitments (which were set by the Labor Federal Government) without launching these two satellites. I also believe that the company has done extensive due diligence on the matter. Part of this belief is based on what I have seen of NBN Co’s contract processes, which have always appeared to be above board, but another part of it is that I don’t believe an executive of Dawson’s stature would willingly tell porkies about core technology infrastructure and procurement. He has no motivation to, after all — NBN Co is working for the public good of Australia. And it’s a company of engineers, not a company of politicians.

With respect to the broadband policy question, I do believe that remote Australians deserve and will definitely fully utilise broadband speeds of 12Mbps and increased quotas. Certainly we do in the cities — in rural areas such as farms and properties I can only see an increased need for good broadband.

Of course, as always, I will be happy to look at any evidence to the contrary — please send it to me if you have some — but for now, technically speaking, it appears as if NBN Co has made the right technical decision with its satellite contract announcement this week.

Image credit: NBN Co

56 COMMENTS

  1. I agree. Turnbull was a little hasty with his comments about the satellites. In the end, the whole point was for rural users. The Coalition had a $1bn OPEL contract to simply deliver to rural Australians. That’s all we needed. That would have used terrestrial wireless and wouldn’t have had complete Australia coverage. I think $620M for rural coverage is a good outcome, perhaps there could have been slightly better contract arrangements, but it’s good.

    Of course, we don’t NEED 100Mbps FTTP to suburban houses. FTTN would have been fine. The whole NBN could have cost the Government $1bn, not $30-$50bn. (in this frame of mind, of servicing the rural areas).

    This could be scaled up even further to 20 satellites (in financial terms), for arguably less than $6bn to cover 10% of the population. And would probably be prudent, given that it will be a little while before commercial terrestrial wireless has that sort of coverage – imagine all those adventurers, miners, flights, remote assets, etc.. which could benefit. And with capacity to grow.

    Don’t hold me to these ideas. They’re just ideas.

    • FTTN doesn’t cost $1 billion, it actually cost (depending on who you talk to, ~$6 billion according to Coalition if you read their broadband policy OR according to Citigroup, ~$17 billion.

      • And you have to enable a monopolist, effectively culling the entire rest of the retail access industry.

    • Because the National Party can’t be seen to be agreeing with the Labor Party, they probably love the idea and have wanted something like this for years but have never been able to get the idea up there for discussion and if they did Labor & Liberal would have never agreed to it. You have to love the politicians never for there electrets always for there party’s.

      • In fact it WAS in the National Party’s original policy platform before the 2010 election, but was later toned down.

        I even wrote this article in July 2010 suggesting what their policy might look like based upon their documentation and coalition recommendations to the NBN Senate Committees:
        Broadband access could be an election issue, 24 July 2010
        http://www.newsweekly.com.au/article.php?id=4374

        But then Tony Abbott promised – 6 days before the poll – to deliver more of the same and no Telstra separation, losing three hitherto safe regional seats and what should have been an unloseable election.

  2. Well, is the satellite network going to serve 200,000, “several hundred thousand” or “several hundred” as these numbers are used in the article? Why are we still condemming a large portion of the Australian public with sub-standard communications when it has been costed and proven that fibre can be extended out to rural and remote communities in a timely fashion, and reasonable price.

    The web experience for satellite users is very poor, apart from watching movies, for which it is well suited. Anything that uses a Web 2.x paradigm or AJAX transactions is going to be painful to use. Even CSS and to the point, CSS3 web applications, so lovingly crafted over fixed line and wireless, are going to give a very poor result in the satellite world. What happens to an onMouseover or onHover event when some 20-200 HTML objects are sending information to the application when the latency is around 1 second? You are going to have to drag that mouse around pretty slowly to see the application working. What about keyboard events? Same deal. Regardless of how sexy a $2 BILLION dollars shopping spree looks, a satellite is just a repeater. Nothing more, nothing less. And the geosynchronous repeaters are parked quite a way away. Those of us that have been involved in the design and deployment of satellite systems understand the limitations of this communications media, and it is rather surprising that a ‘nation building’ mega spend is going to leave our mates in the bush with a poor solution. A mouse event on your iPod traveles from the iPod to your house wireless router, then to the NBN terminal equipment, then SHOOSH! via satellite 35,786 km UP to the satellite, then 35,786km DOWN to the application provider (and a few networks, routers, switches come in play during this level of the hop), the AJAX transaction returns the result of ‘SELECT account_name FROM clients WHERE account_name LIKE %ADD%” back UP 35,786km then back DOWN to Joe in the bush, some 35,786km away. That’s 143,144 km not counting terrestrial hops. The latency is a killer, regardless how new the space-based repeater. Several techniques have been investigated and invented to help overcome this problem, application engineering, modified TCP-IP stacks (SCPS-TP), and user education in exactly what to expect from satellite communications services. TV looks fine over satellite. Once the first chunk-O-data has been recieved, stream away. TCP/IP has a limitation in the packet size being sent, the ACK request, NACK re-transmission, packet ordering and the TCP “slowdown” that occurs when packet loss reaches limits. New applications are “chatty” with every HTML object having a potential communication/conversation with the application. And as the “chatty” transactions are usually quite small, more overhead is generated for the network to support. This is reality. NBN groupies are fond of mentioning Physics, well here it is. Latency sucks. People wanting to play games or use Web 2.x services are going to be left out (again).
    There is an assumption in the article that anyone who is not an NBN groupie, is likely to argue that those out bush do not really NEED fast internet. I don’t know your social circles, but for myself and my peers, the opposite is true. I argue against the NBNCo model as it is NOT providing suitable services to remote users. What is the alternative? This has been modelled and costed in the past. Building the NBN on a FTTN architecture (using FTTP where required) provides a homogeneous network diagram. The main high-speed trunks are run out to node using existing fibre in the ground, with extensions and expansion as required. During 1999-2000 I build satellite Earth stations in remote Arnhem Land in aboriginal communites because that network architecture was the only choice. Recently, Telstra, the NT government and Rio have laid OFT into the same areas providing FTTN (and some FTTP) connectivity into these VERY remote locations. All the way over to Groote Island now.

    “Arnhem Land connects to high speed broadband
    On 2 December 2009, Telstra announced the completion of one of the largest optic fibre and broadband
    infrastructure projects undertaken in recent times – the Arnhem Land Fibre Project.
    The $34 million project, which received funding from the Northern Territory Government and Rio Tinto
    Alcan and valuable support from the Northern Land Council, connects nine Indigenous communities and the township of Nhulunbuy to the nation’s fibre optic backbone. Stage 1 of the project (completed November 2008): incorporated laying over 800km of fibre optic cable between Jabiru and Nhulunbuy and building associated telecommunications infrastructure to provide access to world class, high speed broadband. Stage 2 of the project (completed November 2009): required a further 190km of fibre and five radio systems to each of the islands. Due to the project, high-speed broadband and services, equivalent to those in our largest cities, are now accessible to some of Australia’s most remote communities – connecting approximately 10,000 people. The project, which was completed on time and on budget, saw over 800km of fibre optic cable laid across fragile terrain, in difficult climatic conditions between Jabiru and Nhulunbuy.”

    Seems strange the NBN wants to spend another $2 BILLION dollars to provide satellite broadband to:

    1. Remote communities that now have Fibre
    2. Remote communities that already have satellite broadband.

    Telstra prepared a detailed case during 2009, and as part of the executive summary stated:

    “Telstra estimates $250 million in funding, if allocated solely to extension rather than duplication, could connect 140 communities in Australia to the high- capacity, national transmission networks, removing barriers to deployment of high speed fixed and mobile communications. Some of these communities are the most remote and isolated in Australia. ”

    The report documented these 140 areas as the rural and remote communications ‘blackspots’ in Australia and had a plan that would bring these locations high-speed LOW LATENCY network connectivity for a $250 MILLION investment.

    And the model was proven to work on the Arnhem Land Project.

    Full report here:
    http://www.addinall.net/nbn/Telstra-2.pdf

    ‘Future Proofing”. We are all aware I take it that the life expectancy for a modern geosynchronous satellite is in the order of 15-18 years right? If we launch in 2015 they are the way back down by 2030-33. What then?

    Ka band communications. I find it somewhat surprising that during the extensive research on satellite communications in Australia and worldwide that Ku band didn’t get a mention. Never mind. Ka band is the new(ish) kid on the block.. It has however some very real problems when it comes to rain and dust

    High frequency bands like Ka-band are starting to be used in more often in satellite communications networks. The Ka band (30/20 GHz), however, the effect of or dust rain attenuation is more significant than in lower frequency bands, such as the popular Ku band (14112 GHz) and Charlie band.
    Some countermeasures for rain attenuation using site and/or satellite diversities are required in order to operate reliable Ka and satellite communication links. In a human sense this means over-engineering of the transmit/receive stations. If the network topology followed the homogeneous FTTN pattern, then large Earth stations would be built replacing the Fibre portion of the network feed with the satellite system. Over-engineering (dish size, computational power, energy requirements etc) can be made at a node. This is less likely when the satellite system is conversing with a home installation.

    Given a decent network architecture that is FTTN at a national level, whilst still being able to use ‘best fit’ technologies in individual circumstances would reduce the number of people relying on satellite communications from “several hundred thousand” to perhaps 100,000 or less. These poor folk can be serviced by purchasing bandwidth from new space missions such as Jabiru-2 and other sophisticated Ka Band birds soon to fly, removing the requirement of the NBNCo to be the owner of space missions.

    I really feel that the money spent on this portion of the NBN model is not money well spent, and the network is poorly planned.

    Cheers,
    Mark Addinall.

    • Hey Mark,
      that would be a lot easier to read with some basic formatting like paragraphs. ;)

      I knew satellite internet wasn’t fantastic but I never really thought about how much latency is involved.

      • Sorry Martin. It resolves into paragraphs on this machine (Fedora 16, Firefox 10).

        Latency sucks, not only for state of the art Web 2.0 applications, but also for the older apps. One application that needed to go on the satellite network was designed for Wyse30 terminals, being driven by a terminal emulator at 9600 baud.

        One would think that such a lightweight system would be OK on the satellite net. Not so. As was common 15-20-30 years ago, each character press was transmitted to the application server, tested to see if a “special key” (ie. F1 – HELP) had been hit, if not, transmit back down for ECHO CHAR.

        That was painful. I re-engineered the transactions to send screen at a time “chunks”. Still a bit horrid.

        Other systems I worked on (www.addinall.org/parakeet.jpg) we were given a free hand to tailor the applications towards a satellite environment.

        The web is going the other way, with chatty and RESTful transactions being the norm.

        Cheers,
        Mark Addinall.

          • Not SEEN Hubert, DID,
            [censored]

            Anyway,
            “Geostationary unsuitable for low-latency applications
            Satellite Internet dish on the side of a house in rural New York State

            A geostationary orbit (or geostationary Earth orbit/GEO) is a geosynchronous orbit directly above the Earth’s equator (0° latitude), with a period equal to the Earth’s rotational period and an orbital eccentricity of approximately zero. An object in a geostationary orbit appears motionless, at a fixed position in the sky, to ground observers. Communications satellites and weather satellites are often given geostationary orbits, so that the satellite antennas that communicate with them do not have to move to track them, but can be pointed permanently at the position in the sky where they stay. Due to the constant 0° latitude and circularity of geostationary orbits, satellites in GEO differ in location by longitude only.

            The internet latency mentioned above makes satellite Internet service problematic for applications requiring real-time user input, such as online games or remote surgery. This delay can also be irritating and debilitating with interactive applications, such as VoIP, videoconferencing, or other person-to-person communication. It will cause most general market applications (such as Skype) to behave unpredictably and fail, as these are not designed for the difficult compensation required for the high-latency connections. Some people find that the delays inserted into conversation over a high-latency connection make communication difficult and may lead to a feeling of mistrust or hesitation, even when both sides are aware of the lag.

            Latency also impacts the initiation of secure Internet connections such as SSL which require the exchange of numerous pieces of data between web server and web client. Although these pieces of data are small, the multiple round trips involved in the handshake produce long delays compared to other forms of Internet connectivity, as documented by Stephen T. Cobb in a 2011 report published by the Rural Mobile and Broadband Alliance.[3] This annoyance extends to entering and editing data using some Software as a Service or SaaS applications as well as other forms of online work.

            The functionality of live interactive access to a distant computer can be impaired by high latency. While these problems may be tolerable for basic email access and web browsing, the use of character-by-character command shell or virtual private networks (which typically involve several round trips using layered protocols) is almost impossible through geostationary connections. For this reason, the two largest satellite Internet providers in North America, HughesNet and WildBlue, do not recommend their services be used for VPNs. The HughesNet FAQ states: “Virtual Private Networks do not work well over satellite…HughesNet Technical Support does not provide help with…problems associated with VPN clients.”[4]”

            From Wiki. That’s $2 BILLION to provide an unusable internet connection. Good plan _Hubert_!

            Seems these forums are just for kiddies. Pity.

            Mark Addinall.

          • Not SEEN Hubert, DID,
            [censored]

            I take umbrage at you censoring my work. “Abo’s” is an English contraction of the word ‘Aborigines’. Contractions in English are common and more so in the Australian idomatic English usage.

            “A contraction is a word or phrase that’s (or that has) been shortened by dropping one or more letters. In writing, an apostrophe takes the place of the missing letters.

            We rely on contractions all the time in normal conversations. As Ben Yagoda says in The Sound on the Page (Harper, 2004), “In speech, there is an expectation that anyone who’s not prissy or pretentious or is emphasizing a point will use [contractions] whenever possible.”

            Some people are under the impression that contractions should never appear in writing, but this belief is mistaken. The use of contractions is directly related to tone. In informal writing (from text messages and blogs to memos and personal essays), we often rely on contractions to maintain a colloquial tone.”

            It is in common usage by Yolŋu and Larrakia language groups in the Pisin of everday language.

            I leave you to bask in the glory of the script kiddies you use for a captive audience.

            Mark Addinall.

        • “As was common 15-20-30 years ago, each character press was transmitted to the application server, tested to see if a “special key” (ie. F1 – HELP) had been hit, if not, transmit back down for ECHO CHAR.

          That was painful. I re-engineered the transactions to send screen at a time “chunks”. Still a bit horrid.”

          I had the same experience back in 86. One of our customers decided rather than use ISDN and a multiplexer between his Melbourne and Sydney offices he could save some bucks and use the internet. I think it was AARNET. Anyway, they were, I think Wyse 50s at the time, not sure. Usual thing, sends a character, echos it back. Trouble is the charging system at the time was on packets, can’t remember the exact size I think they were likely about 1.5K. Echoing single characters was not good, not to mention the latency as it timed out on getting extra data for the packet then sending it. As you had to, it meant a lot of work rewritting the software to send blocks of fields from the terminal. Because of this they also lost all the fancy features (at the time) like autocomplete, etc.

          • Jesus! Swapping from It Still Does Nothing over to AARNET in 86 must have cost a penny. Mine in 89 cost around $64,000 (magic number). I have been hitting the ECHO CHAR thing for ages. Just as I joined Sybiz in 84 a bloke hired me to hook up his KayproII via acoutic coupler to a 3174 Terminal server! Eeeeekkkk!

            Mark Addinall.

          • I don’t know the figures on the cost. I have this feeling it was way more than the guy was expecting. We’d specified 64Kb ISDN for the 6 terminals hooked up to a multiplexor. I can’t even remember what that cost, I think it was like 25-30K
            I think someone from AARNET knew the system administrator. He’d done some calculations based on the amount of data, etc, but it was all done with full packets. I don’t think it was that much cheaper 20K. But it was the future and that’s what they put in. We didn’t know til we got there. We also had no idea how they were getting charged, just that they had 6 terminals hooked up and it was slow and laggy.
            It was a month or so latter when the shit hit that fan. I think the first bill was way of 50K
            They still continued with it though, paid to have the software modified. And it wasn’t a small job, it was an entire integrated business system, sales, stock control, general ledger, payroll, manufacturing. I still think it was costing them a fortune, but it was their call.

    • “Given a decent network architecture that is FTTN at a national level, whilst still being able to use ‘best fit’ technologies in individual circumstances would reduce the number of people relying on satellite communications from “several hundred thousand” to perhaps 100,000 or less”

      I guess I’m confused…are you suggesting that a FTTN network is in any way a solution for low density areas? In what way would FTTN be a better solution going forward than a FTTP solution?

      “We are all aware I take it that the life expectancy for a modern geosynchronous satellite is in the order of 15-18 years right? If we launch in 2015 they are the way back down by 2030-33. What then?”

      I expect that by that point there will be at least 1 more FTTP rollout…if you figure that our population will have increased by at least 50% (based on the average), then obviously the rollout will be an ongoing program to keep up with the country and where people end up living. It certainly won’t be like the HFC rollout (one shot and that’s it).

      • Indeed Chas.

        I believe that by 2034, NBN Co have forecast that the NBN will have been repaid.

        So mission accomplished.

      • ““Given a decent network architecture that is FTTN at a national level, whilst still being able to use ‘best fit’ technologies in individual circumstances would reduce the number of people relying on satellite communications from “several hundred thousand” to perhaps 100,000 or less””

        “I guess I’m confused…”

        I am not surprised. Hardly any technical debate has been allowed on the architecture or topology(ies) of the NBN. The proposed NBN was thought up over lunch and given to NBNCo to implement. Due to the abandonment of quite well documented practices and procedures for government ICT procurement and programmes, the current NBN model is a mess, and not likely to improve. Why are NBNCo suggesting that they want to LIFT the numbers of people on satellite communications when it is well known that it is a poor choice for modern web usage? The only reason I can see is that it provides an ex post facto justification for another unmanaged $2 BILLION spend-up.

        “are you suggesting that a FTTN network is in any way a solution for low density areas? In what way would FTTN be a better solution going forward than a FTTP solution?”

        In both cases, yes. If one takes Australian civilian public internet communications as a whole of country infrastructure, then a more homogenous network architecture could have been had by assuming that fast backbone will supply NODES. One must consider a node as just a distribution and collection point for internet data. An example of a node in densly populated areas can be FTTC (be this Fibre to the Cabinet or Fibre to the Curb) or FTTB (Fibre to the basement, a model in-between FTTH and FTTC, in reality, FTTN). I currently run FTTH installed by Telstra a few weeks ago, although in reality it is a FTTN/FTTB installation as my wireless router that supplies my machines connects to the NODE in the basement via CAT5 copper. So once the political spin has been removed from the argument over design, the engineering can get on with designing a suitable solution. Where the NBN model is wrong, is that it concentrates on providing FTTH to everyone they can get to in the cities and larger towns, and leave the under-supported population still under-supported. Dwellings in the densely populated areas are getting FTTP whether they WANT or NEED it. A model based on FTTN with differing granularities of what construes a “NODE” can be designed so that the architectural model remains consistant regardless of geographic location.

        In the cities, the last mile technology can be chosen as a connection to the node, be it true FTTH, LTE-Advanced, VDSL, ADSL2+ or “bugger off, we don’t want no steenking internet”.

        High speed fat OFT pipes will connect these nodes. Without question, and to the greater extent in Australia ALREADY DO. The plan put forward by Telstra was to extend the OFT that is already in the ground to enable such a model. This type of engineering was carried out on the Fibre to Arnhem Land project, where OFT communications were taken all the way into remote Arnhem land providing FTTN OFT that drives high speed LTE, and in cases where FTTH has been requested (schools, medical centres etc) it has been provided.

        There are places in Australia where you just can not roll out a fibre backbone and make it cost effective. In this instance, the LTE NODE(s) can be supplied bandwidth by point-beam focussed high speed Ka satellite. The latency is still there, no getting away from it, however the ideal for the nation would be to DECREASE the number of people relying on satellite communications, not INCREASING the percentage.

        The govefrnment can easily make the soon to be defunct 700MHz analogue television spectrum a government owned asset, and internet broadband can be carried on these frequencies to very remote locations.

        The few that are left with no choice other than a direct satellite link (removed from the FTTN architecture) can be serviced by leasing space on existing or planned space missions.

        “We are all aware I take it that the life expectancy for a modern geosynchronous satellite is in the order of 15-18 years right? If we launch in 2015 they are the way back down by 2030-33. What then?”

        “I expect that by that point there will be at least 1 more FTTP rollout…”

        That does not have any bearing on the fact that the two new satellites will have entered end-of-service lifetime. By the time the NBN is complete (if left to go on in this fashion) the brand new satellites are junk metal. So you NEED two more. Or then START building a decent FTTN architecture out to remote and rural Australia.

        “if you figure that our population will have increased by at least 50% (based on the average), then ”

        That is well off the mark. I did Demographic software for the ABS. Easy to look up.
        By June 2026 the ABS predicts that the population of Australia will be between 25 972 and 28 723 million peoples (not households)

        http://www.abs.gov.au/Ausstats/abs@.nsf/mf/3222.0

        The current population is around 22,829,300 so that makes a population increase of 13.77% to 25.81% in the time frame we talk about.

        “obviously the rollout will be an ongoing program to keep up with the country and where people end up living. It certainly won’t be like the HFC rollout (one shot and that’s it).”

        So the NBN NEVER has an end date? That makes it a nationalised service, and so, needs to be added to general revenue expenditure.

        The model is poor. Increased reliance on satellite systems make the model poorer still.

        Mark Addinall.

        • “In both cases, yes”

          Makes no sense at all…by using Nodes in a low density area, you get very few connections per Node (due to the deterioration of signal at a distance from the Node). That’s why NZ is changing their FTTN to FTTP as well…FTTN is far too inefficient. The only place it makes any sense for a short period (very short) might be in high density areas…the higher the better. That allows for a far higher efficiency for each node.
          That said, using FTTN is merely putting off the inevitable at a long-term cost that is unacceptable…

          “One must consider a node as just a distribution and collection point for internet data”

          You forgot the line “with very limited range, relatively speaking”. So your choices for FTTN in low density areas are to either continue with speeds approaching dialup slowness, or build a HUGE number of Nodes…for that money, it makes far more sense to just do it right with FTTP in the first place (if it can be afforded). Otherwise, satellite is far preferable to FTTN at those distances.

        • “By June 2026 the ABS predicts that the population of Australia will be between 25 972 and 28 723 million peoples (not households)”

          Ummm…no. Current growth rate is 1.4%/year…you are figuring the natural growth rate (and that estimate was from 2006/7)…
          Try taking the current population, and adding 1.4% compounded for the next 20 years and see what you get (don’t forget the compunding as this is a function of growth).

          “So the NBN NEVER has an end date? That makes it a nationalised service, and so, needs to be added to general revenue expenditure”

          That is so left field…the answer is no, it never has an “end date”. Why would a GBE have an end date?
          Do other businesses have one?

    • “can you please make your posts shorter and with better paragraphs … you have some interesting and informed info, but your stuff is getting hard to read ;)”

      Shorter is not that easy. It seems I am communicating with people who really do not have a grasp of the basics of network architecture and operation. That is OK. However to explain WHY latency is a BAD thing for a modern application to an audience with limited skills, one has to go back a step further than is usual in a technical forum. Apologies. I spend a great deal of time in front of a whiteboard.

      The paragraphs look OK to me. Getting 50-200 words per paragraph. A bit long for a novel, but pretty average for a technical paper.

      I shall try harder.

      Mark Addinall.

  3. I’m afraid that the Australian Government has already tried the satellite game, in their last attempt at social engineering in telecommunications policy. Assat sunk into $1 billion dollar morass, before being quietly disposed of as part of the licence cost to establish Optus.

    Fundamentally, the Rudd/Conroy policy decision to remediate Howard’s flawed divestiture of Telstra by recreating a 21st Century Post Master General’s Department couldn’t be worse for our aspirations for long term competitiveness. But unfortunately, it will be 2030 before this assertion can be adequately debated.

    So, in the end, we’re saddled with what was voted in. Nothing more can be said.

    • Actually Howard had an even worse policy, it was called OPEL.

      And they are not recreating MGD.

      The difference being with the NBN is that Level 2 Wholesale where NBN just looks after the lastmile.

      It’s also creating a large backbone and core network, which we in Australia, desperately needing.

      • “It’s also creating a large backbone and core network, which we in Australia, desperately needing.”

        We already have a large OFT backbone?/core network. And chucking birds into space is not going to improve it much eh?

        Cheers,
        Mark Addinall.

        • I disagree, alot of the existing fibre can be easily 10 years old, and by the time NBN is completed, will be 20 years old.

          NBN Fibre section will consist of:

          1. 121 POI.
          2. 13 Million premises to be covered.
          3. 181 Million KM’s of gigabit-capable Passive Optical Network.
          4. 57,000km of transit fibre (which I believe is the Transit network of 10Tbps).

          There is also

          – FDA (Fibre Distribution Area)
          – FSAM (Fibre Servicing Area Module)
          – FSA (Fibre Servicing Area)

          Of which the fibre will have to connect too, so these will have to be built.

          The excuse of “chucking birds into space” does not cut it.

          • You sed() … “I disagree, alot of the existing fibre can be easily 10 years old, and by the time NBN is completed, will be 20 years old.”

            I thought this OFT stuff was supposed to last forever? So that means that the fibre laid down today by the NBNCo will need to be ripped up in 2021-2031? Seems a waste of time then really. Although it will co-ncide nicely with the satellites achieving end of service life.

            The numbers look pretty. However, a lot of what the NBNCo proposes is overbuild, which in a teensy country, it is wasteful. In a large sparsely populated country it is madness.

            The proposed network map of NBNCo fibre network is almost an EXACT copy of the Telstra existing OFT network. Even down to some of the private network segments in NNW Australia built in partnership by the miners and Telstra!

            I wonder if I am the only one to notice this?

            “The excuse of “chucking birds into space” does not cut it.”

            Cut what? I don’t get you.

            Cheers,
            Mark Addinall.

  4. When the Coalition were in government and they were considering their own national broadband upgrade, that included two satellites to serve the people who couldn’t be served by terrestrial networks.

    • Hi,

      sure, let’s look at NewSat. According to their website, they have:

      Jabiru-1: 7.6GHz of capacity to the Middle East, Asia and Africa.
      Jabiru-2: 216MHz of capacity across “high demand regions including Pilbara, Kimberley, North West Shelf and Timor Gap”, as well as Timor Leste and Papua New Guinea
      Jabiru-3: 6GHz of capacity over Africa, the Middle East, Asia and Europe (excluding most of Australia)
      Jabiru-4: 6GHz of capacity ” from North East Asia and Australia, across the Pacific Ocean to the West Coast of the USA”
      Jabiru-5: 6Ghz of capacity “over Central, North and South America, as well as Western Africa and the Atlantic Ocean”

      In comparison, NBN Co’s two satellites will deliver 90GHz of capacity across Australia.

      I’m sure NewSat is a fine company with many great solutions. But what NBN Co is planning is on a different scale to what NewSat can offer. Have you looked at the technical details of their satellites? They’re all available on their website.

      This is a technical discussion — let’s stick to the facts here ;)

      Cheers,

      Renai

  5. Mark Addinall says that “costed and proven that fibre can be extended out to rural and remote communities in a timely fashion, and reasonable price.”

    Satellite broadband is required for the remote farms and stations. Many of these have only one family living on the property. It will never be economic to plough fibre to places like these. Nor would your suggestion of a satellite version of the FTTN be practical for a single family on a large pastoral property.

    Note that we are not just talking about the far north here. Satellite broadband is the option used beyond the reach of NBN Co’s LTE wireless cells and most users will be in the south west and south east of the country.

    • “Mark Addinall says that “costed and proven that fibre can be extended out to rural and remote communities in a timely fashion, and reasonable price.””

      Indeed I did. Telstra just took FTTN and FTTP out to remote Arnhem land for a budgeted and real cost of $34 MILLION. Using the Arnhem land project as an engineering baseline Telstra said it could cover the known 140 Australia wide broadband blackspots for an investment of $250 MILLION.

      “Satellite broadband is required for the remote farms and stations. Many of these have only one family living on the property. It will never be economic to plough fibre to places like these.

      This is indeed true.

      “Nor would your suggestion of a satellite version of the FTTN be practical for a single family on a large pastoral property.”

      There are going to be some families left out in the cold, where the only option is satellite. But an INCREASE from a current 100,000 (projected) users to a fictional doubling to 200,000++! This means that the NBN is planning to make connectivity actually worse for a great number of Australians.

      The FTTN architecture works well in the VERY few spots one can’t run an OFT pipe. Given the analogue television spectrum is soon to be in the disused bucket, Ka band spot feeds to Nodes that transmit LTE MIMO of the 700MHz spectrum would cover the remainder of the hard to get at places. The people living on HUGE ranches in Australia do not number in the hundreds of thousands.

      “Note that we are not just talking about the far north here. Satellite broadband is the option used beyond the reach of NBN Co’s LTE wireless cells and most users will be in the south west and south east of the country.”

      As I said. Poor network planning.

      Mark Addinall.

      • You realise it was $34 million for the link.. which replaced the previous microwave link. It has nothing to do with FTTN or FFTP. The average punter is still using ADSL. It just means that most importantly the government services (especially to the hospital) don’t go down during a storm which would happen all the time during the wet season.

        • “You realise it was $34 million for the link.. which replaced the previous microwave link. It has nothing to do with FTTN or FFTP. The average punter is still using ADSL. It just means that most importantly the government services (especially to the hospital) don’t go down during a storm which would happen all the time during the wet season.”

          Sorry Matt but WHAT are you talking about?

          Mark Addinall.

  6. For the stated task, the dimensioning of this system seems about right. Ka band is the only place the necessary spectrum can be found, however Ka does suck when it rains.

    Broadband that works most of the time is much better than no broadband, but these people might need some other form of communications for emergency phone calls. The 0.2% of Australians in truely remote locations will be accustomed to intermittent comms, but the tree change folks will spit the dummy if they can’t use the phone when there is heavy rain.

  7. Turnbull says there is enough capacity to rent off existing satellites.
    This coming from a man who says that there is enough capacity in copper & HFC networks to serve the population for the foreseeable future.

    I also wonder what hell would there be to pay, should the satellite we lease capacity off – dies. How long before the remote folks get their net back. Where is the redundancy?

    Kudos to NBN Co.

    Think about this…
    I say “Turnbull says”
    I didn’t say “Turnbull believes”
    … I wonder how things would change if Abbott wasn’t pulling his strings.

    • Turnbull’s mandate is to destroy the NBN, oppose whatever Labour does. If Turnbull says Labour should do something and they do exactly what he says they will still have done it wrong.

      What would be wierder is if he said well done, they seem to have made a good decision in this case.
      Think about it, wouldn’t you check to see if that article reporting this was an April fools joke?

      I am sure what he has to say and what he actually believe conflict a lot of the time. Early on, just after Uni, before I moved to Melbourne, living in the country there was not a lot of programming work. So as well as doing contract work, when available I had a few other jobs. Some were sales. I found if I believed in a product I could sell it, if not the job was a chore, and every day you felt lower and lower.
      I could never be a salesman or politition. Working as a professional liar has got to be a sole destroying profession, if in fact you have a soul to begin with. Shudder, just thinking of it makes me wonder about the morality of someone who can lie 24/7 and live with themselves. It’s something I can’t do.

      • Noddy sed() “I am sure what he has to say and what he actually believe conflict a lot of the time. Early on, just after Uni, before I moved to Melbourne, living in the country there was not a lot of programming work. ”

        I worked for BREEZE back in pre-history. At least it was near surf beaches.

        Mark Addinall.

        • Sounds nice. Unfortunately I was nowhere near the coast. All Bendigo had was a lot of petrol heads. At least it was the right time to get some work. Schools were starting to get Apple II systems, a little later Obsbourne and Kaypro and their CPM boxes meant small businesses wanted to automate. Luckily I scored some work repairing computers for local computer stores (all two them) and that lead into some contract programming work. All full time work was more operator stuff. So to get a full time programming job, Melbourne it was. At least now I live right on the beach, though have to drive to the back beaches for surf.

          • “Sounds nice. Unfortunately I was nowhere near the coast. All Bendigo had was a lot of petrol heads. At least it was the right time to get some work. Schools were starting to get Apple II systems, a little later Obsbourne and Kaypro and their CPM boxes meant small businesses wanted to automate. Luckily I scored some work repairing computers for local computer stores (all two them) and that lead into some contract programming work. All full time work was more operator stuff. So to get a full time programming job, Melbourne it was. At least now I live right on the beach, though have to drive to the back beaches for surf.”

            I spent some time coding out Mildura and a bit out at Bendigo Uni. Central Vic is a lot of fun! ;-)
            Still surfing at 60++? Cool. Hope for me then.

            Mark Addinall.

  8. NBN Co is supposably at arm’s length from government, why is that the minister or PM has to make all their announcements or “officially” turn on new infrastructure?

    • Politics. Any successes are the governments. Any failures are NBN Cos.
      From the Coalition point of view, all is failure and all is Labours fault.

  9. Renai, I haven’t actually seen NBNCo say the satellites are geo-stationary, and I have read that KA-Band satellites are often used at lower orbits which obviously allows them to provide better latency.

    Have you actually heard they are geo-stationary from a press release or is it just assumptions.

    I think it would be good to find out exactly how high these satellites will be sitting and what latency users can expect from them, and also during storms etc.

    • It has to be geosynchronous eh? TWO satellites, one for redundancy (what a dumb idea) have to fit into very close slots to enable a broad cell coverage of Australia.

      MEOs and LEOs cut down on latency, however.

      1. You need LOTs more satellites.
      2. Where the latency is better, the speed is poorer. Typically 64kbps – 512kbps.
      3. Two way transmission is difficult, usually requiring a MILSAT following Earth station.
      4. As the orbital height decreases, the orbital decay increases, giving LEOs a shorter life span.

      All in all, satellites are just a BAD idea to deliver next gen broadband services to Australia

      Wiki.

      “For geostationary satellites, there is no way to eliminate latency, but the problem can be somewhat mitigated in Internet communications with TCP acceleration features that shorten the round trip time (RTT) per packet by splitting the feedback loop between the sender and the receiver. Such acceleration features are usually present in recent technology developments embedded in new satellite Internet services.

      Medium Earth orbit (MEO) and low Earth orbit (LEO) satellites do not have such great delays. The current LEO constellations of Globalstar and Iridium satellites have delays of less than 40 ms round trip, but their throughput is less than broadband at 64 kbit/s per channel. The Globalstar constellation orbits 1,420 km above the earth and Iridium orbits at 670 km altitude. The proposed O3b Networks MEO constellation scheduled for deployment in 2010 would orbit at 8,062 km, with RTT latency of approximately 125 ms. The proposed new network is also designed for much higher throughput with links well in excess of 1 Gbit/s (Gigabits per second). The planned COMMStellation™, scheduled for launch in 2015, will orbit the earth at 1,000 km with a latency of approximately 7 ms. This polar orbiting constellation of 78 microsatellites will provide global backhaul with throughput in excess of 1.2 Gbit/s.”

      Mark Addinall.

      • Mark, two redundant satellites is not a dumb idea, but basic risk management.

        Have a read through the technology discussion in the May 2010 NBN Implementation Study.

        The short version is that a single catastrophic failure of one satellite would see large parts of Australia suddenly without a service on which they had come to depend. Months or years needed to build and launch a replacement would constitute an unacceptable outage risk. Consequently you launch two satellites. They could ordinarily operate in a load-sharing mode, but either could take the full traffic load when necessary, e.g. if an orbital fault needed correction and this turned the transponders briefly away from Australia.

  10. “Ouch! That’s got have been some major work. I managed to do one of mine when i was younger. The limit of adhesion of a vans tires going around a corner wasn’t as high as I hoped.”

    My last was the worst. If you go mountain climbing without ropes, don’t let go, it hurts! ;-)

    “My start was a 2650. I remember saving for months to get 80 bucks together to expend it from 1K to 2K.
    Next was the TRS 80. I regreted using it as a trade for a C-64 when they came out. The TRS 80 was much nicer to program. I’d gotten the Pascal compiler and assembler for it.”

    Yeah, I had Pascal and assembler for the Trash 80. Came in handy later in life when I moved into a USCD P-System development house. I kept mine for ages. Welded an S-100 BUS and a 5″ floppy onto the thing. Almost turned into a computer! I wrote a word processor that came in at 7.2KB. FAT FAT FAT.!!

    When we first published Sybiz on MS-DOS, General Ledger, Creditors Ledger, Debtors Ledger, Order Entry/Invoicing, Stock control, Reporting all fit on ONE 360KB floppy!

    Took 1 hour 12 minutes to compile. Good excuse to nip around the corner for a quick Coopers!

    Mark Addinall.

  11. I think they were a competitor to the first company I worked for in Melbourne. Howse Fink and Johnson, Distrib. Though they were mainly doing the ICL Clan at first. Once the IBM AT came out it could handle 4-5 terminals. The 386 say 6-8. It took a while to compile too. I remember when the IBM RT came out, 25 minute compile, wow. Shame it used the ISA bus for the serial I/O. One customer had a band printer and the terminals nearly halted every time it fired up. I left there just as the first 486 came out, I think it was only 16-20MHz atr the time, but it could do 25 minutes too. That RT, what a pain in the butt, it was a stop gap while the AS400 got an operating system… they really dropped the ball on that, I think it was over a year. They had an AS400 sitting there they couldn’t use.
    The fastest thing we used at the time was some Fujitsu thing. The railways had a couple, it completed it’s compile in about 30 seconds, we found out once the terminal caught up. Had these huge RAM disk cabinets. God it was quick. Probably almost as fast as a low end smartphone of today :)

  12. I worked for Sybiz twice. They were my first ‘real’ job and took me on as a somewhat older ‘Junior’ programmer. We had an ethos that Sybiz would run on any bloody thing so as bottom of the food chain coder I got a LOT of jobs porting to the vatious flavours of “IBM Compatible” computers!!! A few nightmarish memories! The HP-150 Touch screen, the Sigma OKI (a machine that could not stand on its own. If you didn’t bolt it to the desk, it would fall on you!), Apricot ‘Smart Pad’, the BLOODY DEC Rainbow!!!
    I left after a year and a bit, but went back as team leader soon after. I led the team on the brand spanking new soon to be released Apple Macintosh! We bought two Apple Lisas ($22,000 each) with Pascal compiler and 68000 assembler. Developed on the Lisa then ported the object deck onto the Mac via serial cable. No compilers available for the Mac at that time. I had an unbound copy of “Inside Mac” complete with pencil annoitations by Steve and Woz. My neice chucked it away as part of Uncle Mark’s old computer junk! %-( I was a Stallion when the 486 was released. THE BEASTS! SCO UNIX…Whoaa…

  13. Here’s the thing.

    a) Where there is no connectivity, satellite is better than – drum roll – nothing. It will help cover blackspots in wireless.

    b) Any idiot, frankly, can see where Wireless and or FTTH is available, that’s what will be used. Otherwise, where neither are available, refer a) above.

    c) FTTN is the only choice when FTTH isn’t on the table; it is an interim solution, nothing more. It’s that simple. Fixed line isn’t going away; there will always be a demand for it. When the choice is rotting copper or Fibre, you pick fibre.

    Great waffling diatribes about how evil satellite is, when there is zero alternative (unless Liberal pundits want to add several billion more to cover the 5% with fibre?) are pointless. If wireless is available, it will be used. If fibre is, then it will be used.

    The Coalition strategy, if it can be called one, is to ensure the ongoing Liberal policy of a) not giving a shit, b) underinvesting, c) lining Telstra pockets and d) when all else fails, it’s the fault of the industry.

    They’re not listening to their own state-level party members whom are increasingly making demands for NBN. They’re also not listening to their Coalition partner whom is very much for fixing regional broadband access, and has been for some time now. Never mind what their constituents want, either!

    Frankly, it’s been well established that commercial enterprise just isn’t up to the job of providing great coverage for fast broadband to the majority of Australia. The ROI doesn’t work at commercial levels of return.

    Whenever Turnbull quotes figures, makes statements regarding how commercial enterprise will save the day, he’s proven wrong time and time again. And this is the guy people expect to improve on the NBN?

    Please. Look at the history of the Coalition versus broadband investments. That’s all you need to forecast the future of broadband under any Coalition government.

  14. What people seem to forget is that FTTN is here now, it goes to RIMs, and the last mile is ADSL or ADSL2 if your a Telstra customer. FTTH(standard) allows 4 ISPs to have a presence at your home. The NBN FTTN model was basically similar to the current Telstra fibre to RIM model with ADSL2+ for the last mile, ie 1 ISP at a time, Thank You. And rather like the UK model, you could pay (a lot) extra to have fibre to your house, frankly ADSL is sooo 2000’s.

  15. Only think I learnt from all his ramblings is not to hire mark to code website used outside an internal LAN.

Comments are closed.