• You’re never going to need more than 4 Mbit/s (or so) so there’s really no point in building fiber network.
• You’re never going to need more than whatever LTE can offer because mobile networks are the future.
• You’re going to need more and more capacity so you might as well build fiber-based networks.

Well. It turns out there are quite a few areas in rural Norway that built fiber networks many years ago. So we went to talk to them. During January and February we conducted in-depth interviews with fiber operators, public officials and end users in six rural counties. Here’s what they told us:

1. Fiber networks make a big difference for private enterprise and employment

• Fiber networks are mission critical for businesses that transfer large amounts of data (such as pics and video) to customers and partners.
• Access to fibernets is an important driver for the establishment of knowledge-intense companies with well-paid and interesting jobs.
• Fibernets make videoconferencing easier and more reliable. This is especially important in rural areas where the distance to customers and partners is high.

2. Fiber networks increase quality of life in rural areas

• Fibernets help create more attractive communities. A number of rural areas struggle with an aging population. Fibernets make it easier for young people to move back home after getting their degrees.
• Fibernets improves Internet speeds and TV-offerings. This is especially important for younger people.
• For people with knowledge jobs, access to fiber make it easier to have a home office.  (And yes, for many people a regular broadband connection will suffice, but keep in mind that in rural areas a “regular” broadband connection is usually quite a bit slower than in urban areas.)
• Fibernets are future-proof, meaning that people in rural areas can access future services that are likely to be video-based, such as eHealth and eEducation.  A good example is Bykle county in a mountainous part of South Norway. It is home to a large downhill ski center. Every year, a number of skiers fall and hurt themselves. In order to figure out whether a bone is broken or not, you need to take an X-ray. Before fiber, the skier was sent to the nearest hospital some three hours away in Kristiansand. Very costly and time-consuming. After fiber, the county installed an X-ray machine and had the X-rays transferred to the hospital for analysis using the fiber network. Very smart.

3. Fiber networks improve public administration.

Rural counties tend to have a small population. By linking counties with fiber networks, it is possible to gain economies of scale. For example, three counties in East Norway (Numedal) merged their accounts departments and reduced the headcount by 30%. By creating professional networks across counties, public employees can specialize which makes it easier to attract high-caliber employees.

Readers who know Norwegian can access the whole report here.

The Swedish state-owned company Teracom operates a distribution network for Radio and TV services that covers 99,8% of Swedish households. It’s a nice little earner – the 2008 net profit margin was 20% on revenues of around SEK 3 billion. But some ten years ago the company decided to build a DSL network in 49 Swedish counties with 600 000 homes passed. And that, as this article shows, was not a spectacular success.

The company secured more than €100 million in public financing from the state, the EU and local authorities. In addition, the company invested €40 million, so that the entire network cost almost €150 million. Last September, Teracom decided to sell the network and some 40,000 customers. The selling price? One Swedish krona. Or ten eurocent, if you wish.

Here’s what I think about it:

* Investors no longer value DSL networks (at least not in the Nordic countries). Although the Teracom DSL network valuation is rather extreme, other DSL networks have not been treated well in recent transactions: When Tele2 sold their Norwegian DSL network to Telia last year, the price was around €150 per customer.

* Customers no longer value DSL networks. At least not when we’re talking residential customers and they’re offered FTTH or HFC-based Internet Access.  As these operators continue to upgrade the available speeds customers are likely to migrate away from DSL networks. And in a couple of years even more DSL customers are likely to migrate to new mobile networks when they can offer meaningful and consistent speeds of 10 Mbit/s or so.

Updated in 2010, 12 years later, Nielsen’s broadband capacity observations still follow the exponential growth curve.

Jacob Nielsen's broadband capacity observations

Up until today, three access technologies have competed for the top capacity position.

• xDSL, technologies using telephone lines
• HFC, using former cable-TV networks
• FTTH, using dedicated or shared fiber networks all the way to the customer

As competition for customers has increased, operators have had several public fights over which technology is better or more future-proof. On the 16^th of February, the Norwegian FTTH operator Lyse announced a symmetrical 400 Mbit/s service.  Did FTTH win the race?

To answer this question, let’s have a look at the bottlenecks of the three technologies:

• DSL: The potential of further development of DSL is limited, in our opinion. DSL capacity is limited by the length of the telephone line from the user to the central. The different DSL technologies developed during the last 10-15 years has not changed this, and the major improvements like ADSL 2+ and VDSL primarily helps the shortest line lengths. This means that users very close to telephone centrals may be offered service capacities of 100 Mbit/s. The average Norwegian user, however, may expect to be offered a downstream capacity of maximum 20-30 Mbits/s.
• HFC: The available bandwidth of the coax access part of HFC networks is high, but the topology of these networks make users on the same access segment share it. The typical available downstream bandwidth of ~800 MHz gives a potential of ~5,5 Gbit/s shared downstream capacity with the latest standard, EuroDOCSIS 3.0. This capacity normally is shared among 100 to 250 users. In Norway, this has translated to services promising up to 70 Mbit/s downstream.
• FTTH PON: Passive optical networks, as built by the incumbent Telenor in Norway has the same topology as HFC networks. The main difference from HFC is the access cable, which is fiber rather than coax. The bandwidth and capacity of this fiber cable practically is unlimited. This make the equipment in both ends the bottleneck. In the current GPON implementations 2,5 Gbit/s of downstream capacity typically is shared among 32 to 64 users. Telenor is offering a maximum capacity of 25 Mbit/s downstream today. This may increase as end user equipment is improved, but that is not likely to happen in the near future.
• FTTH P2P: In the point-to-point FTTH networks often preferred by utilities, each user has a dedicated fiber cable to the operator central. This means that the equipment is the bottle neck here as well. Up until now, the interfaces of the operator central and the end user equipment has been limited to 100 Mbit/s. This capacity is shared between broadband and other services, like TV and IP telephony. In Norway, this has lead to broadband service offerings of 50 Mbit/s. Altibox recently announced a symmetrical 400 Mbit/s service to be launched within a few months. Similar services, with up to 1 Gbit/s capacity, may be found in Sweden. The switch from 100 Mbit/s to Gbit/s interfaces in the central and user equipment has made this possible. As for PONs, the interfaces and equipment may be replaced in the future, with 10 Gbit/s per user as the next natural step.

So, did FTTH win? Well, the race between different access technologies will continue, maybe forever. But FTTH has taken a solid lead with up to 1 Gbit/s symmetrical capacity, and will not be passed by any other technology any time soon. HFC will follow nicely behind with downstream capacities of several hundred Mbits/s.  DSL, however, more or less is parked and will be challenged by other technologies, with LTE as an example.

But a few weeks ago I came across a piece (Norwegian only) that I first found unbelievable and afterwards disappointing. Here’s the story: During the last couple of years, a new Digital Terrestrial Television network has been deployed across Norway. This network covers around 98% of the population. Households not covered by this network get a low-cost satellite subscription. But then there are some 5 200 households that cannot access any of these networks. So it was decided to build a “satellite shadow network” of 500 base stations to cover these people. The cost is 1 billion Norwegian kroner which works out to €23 500 or $32 500 per household. Whoa.

Let’s take a look at what you get for €23 500 per household:

• A one way network
• No internet, no telephony
• Three TV channels and 13 radio stations – all from Norway’s public broadcaster NRK

During the last years, we have worked with the rollout of fiber networks in some areas that could be described as: A most rural area in a very rural county in a rural province in a country that is mostly fjords and forests to begin with. And guess what – the deployment cost per subscriber in these projects hover around €10 000. Expensive? Christ – yes, but doable given the right circumstances (something which I hope to blog about next month).

And what do you get?

• A fiber-based two-way network with nearly unlimited capacity
• Up to 50 Mbit/s symmetric internet speed, easily upgradeable
• IP telephony
• 150 TV channels and video-on-demand services

And for €23 500 per sub I am convinced that we could’ve built all of the above and possibly thrown in a redundant connection as well.

The moral of the story?  Well – I think there are several:

• Media organizations, such as NRK, the Norwegian Public Broadcaster who is apparently funding the shadow network using public license revenues, should be very careful when making investments in telecom networks. Building and running a network is very different from their core competency in content creation and management. NRK plans to lay off 100 employees during 2010 due to higher costs than expected. No wonder given their distribution costs.
• Telecom networks built for one specific purpose tend to be hugely expensive and deliver limited value for money.

I spend most of my time advising telecom operators, but we also work for regulators and various government entities. Last year we were asked by the The Norwegian Ministry of Administration and Reform (who is responsible for formulation of the Norwegian broadband policy) to look at the following questions:

1. What is the broadband coverage today?

2. What will future capacity needs look like?

3. What will the rollout costs for the new networks be?

The answer to questions 1 and 3 depends on your definition of what broadband is. Today, the official policy is that everyone should have access to a broadband line with a speed of 640 Kbit/s. And clearly, that is not going to enough when looking a few years into the future. James Enck recently co-wrote an insightful paper which (among plenty of other things which I hope to blog about next month) looked at the need for energy. When the first electricity networks were built, there were few uses for electricity (think light bulbs). But over time the energy networks changed our lives in lots of ways and were possibly the single most important driver for growth during the last century. And the analogy to capacity needs in broadband networks is intuitive and, I think, true. My own favorite broadband definition is that “Broadband is three times your current capacity”.

We built two scenarios for future capacity needs: One primarily based on mobile broadband (LTE) capable of 8 Mbit/s downstream and 1 Mbit/s upstream per customer, and another primarily based on fibre networks capable of more than 50 Mbit/s downstream and 10 Mbit/s upstream per customer (And you may notice that upgraded CATV networks also may be able to deliver 50/10 Mbit/s. For areas that are not covered by fiber or CATV networks, I think it is most likely that they will eventually be covered by fiber networks.) The biggest difference between the two scenarios is video distribution. The LTE scenario depends on a separate network (e.g. satellite, digital terrestrial) for the delivery of broadcast TV.  The fiber scenario can cope just fine with video also.

The rollout costs depend on both capacity needs and government policy. Full coverage with 50 Mbit/s for the 30% or so of households without commercial coverage will cost between two and three billion Euros. We calculated the cost per installed customer to be between €5000 and €7500. This is much higher that other estimates we have seen, but keep in mind that this is the cost for covering rural households in a country filled with mountains and fjords….. In order to arrive at the estimate we simulated a network rollout in 15 counties (with quite different topologies and demographics) and discussed the estimates with installers and fiber network operators.

An important variable for the cost estimates is the availability of ducts and trenches – a government policy where a fiber rollout is planned for when digging for other reasons (such as electricity or water pipes) will greatly reduce rollout costs.

8 Mbit/s mobile broadband coverage cannot deliver all the services that fiber access can, but will cost much less. We calculated to cost to be between €100 and €260 million. (But keep in mind that this is just the cost for covering the 10% or so that will not have commercial coverange). The cost varies with the radio frequency used. In order to minimize the cost it is important to use low radio frequencies with good coverage capabilities.  And here’s the good news: The Norwegian government finally decided last fall to allocate an attractive slice of radio spectrum (the so called Digital Dividend spectrum) for mobile broadband use. Well done.

If you read Norwegian, you can download the report here.