Rising mobile offload to Wifi - saviour of the case for FTTH, or warning sign?

The ever-interesting Benoît Felten has a post up about WiFi offload, with some stats from Mobidia. These figures show that between two-thirds and three-quarters of traffic to mobile devices travels by WiFi, not macrocellular.

His conclusion is:

   Take that, “we’ll only need mobile networks in the future” posse…

And in that he's surely right. No matter how exciting the potential of mobile data, its simply not, at a nationwide level, a substitute for fixed broadband. If the traffic off-loaded to WiFi and the fixed network today were to be 'on-loaded' on to the cellular network, the cellular network would fall over. (This is different of course from decisions at an individual consumer level - some households may decide they can get by with mobile broadband and no fixed broadband).

However, while the future of fixed broadband seems secure, I think usage of mobile devices has a more subtle message regarding the prospects for superfast broadband. Sandvine have recently released their fascinating Global Internet Phenomena Report. One finding is that in North America, 25% of streaming audio and video traffic is delivered to a mobile device in the home - up, presumably, from 0% just a few years ago in the era before smartphones and tablets. This is roughly consistent with figures from the BBC regarding usage of their iPlayer on-demand TV service - 30% of requests for programmes are coming from mobiles and tablets (up from 15% a year ago).

What's this got to do with FTTH? Well, one of the supposed drivers of the need for FTTH is 4K TVs -here's NBN Co getting excited about 85" TV screens. However, if usage of on-demand TV is already shifting to small, handheld devices, that suggests usage on mega-TVs may not be quite what the enthusiasts hope. Mobile devices offer the convenience of a personal device that they can watch in any room of the house (or perhaps in bed - iPlayer's peak of requests is after 10pm). Consumers increasingly seem to be choosing this over watching the content on a bigger screen, be that a TV or a PC monitor.

This isn't an absolute, of course. Households may want the huge screen for the big film on a Saturday night. But for day-to-day use, convenience may trump resolution.


[A footnote : I suspect the Mobidia numbers may be somewhat too high. The data is drawn from the users of Mobidia's 'My Data Manager' app, tag-line: "Take control of your mobile data". I suspect this may not be a representative sample, since such users may be a little more inclined than average to ... err ... take control of their mobile data. They perhaps are more diligent than the average in offloading. However, there's no doubt that WiFi offload is very significant, and drives an ongoing need for widespread fixed broadband.]

Are there any serious analysts still gung-ho for fibre?

Once upon a time it was a lonely path being a fibre sceptic. These days I am in very good company indeed. Here's what leading international telecoms analysts have to say about FTTH:

---

"The disproportionately high spend on [FTTH] is highly problematic. It may come to be seen as inappropriate use of capital in the emerging competitive environment." [March 2012] "The time required to roll out and install FTTH is as large a barrier as cost … There are very few plausible combinations of home services that will require over 100Mbps bandwidth by 2017." [April 2012] "FTTH deployment costs are about five times greater than FTTC costs." [March 2013]

---

"In the current economic climate, it seems unnecessarily dogmatic to espouse ubiquitous FTTH, particularly when the broader ecosystem and regulatory support is out of step with market realities ... [A]lthough deeper penetration of fiber into the network is a must, universal FTTH is an impractical luxury that telcos cannot really afford. … It is difficult for a significant volume of users to justify paying a premium for higher speeds when the applications that they currently use function sufficiently well over high-speed DSL lines" [February 2013]

---

"The combination of sunk costs and highly uncertain demand (both in terms of take up and willingness to pay for ultra-fast broadband services) makes [FTTH] investments very risky … services that would make full use of the higher bandwidth of FTTH are not at present available" [Summer 2012]

---

"The initial focus of the European institutions and of national governments to date has been largely on deployment of fibre-based NGA – [FTTH] – largely to the exclusion of other high speed broadband capable infrastructure. This focus was arguably excessive … More recent statements by the European Commission suggest an increasing recognition of the need for a … strategy that acknowledges the potentially complementary role of other technologies." [September 2012]

---

"Operators’ interest in pay-as-you-grow strategies is also motivated by a dawning realisation of quite how hard – and expensive – rolling fiber right to the home is. In particular, the challenge of installing fibre in front gardens, buildings and individual homes has been “vastly underestimated”" [November 2012] "Australia’s current FTTH-led model … is actually going very much against the global trend of operators using existing network assets to avoid the huge costs of FTTH in brown field sites." [March 2013]

Japan - FTTP with tumbleweeds

Japan is often cited as one of the world leaders for superfast broadband. It has been rolling out fibre-to-the-premises for more than a decade, and now has one of the fastest fixed broadband networks in the world, second only to South Korea in terms of measured speed, with average speeds clocking in at 10.8 Mbps. (The speed of the last mile will be appreciably higher than this).

Of course, we all know that by itself FTTP is just glass in the ground. What matters is what you do with it. How much traffic is travelling across this network? Happily, Japan is one of the countries that tracks and reports this, so we can take a look. [Health warning - what follows is in part based on Google Translate, so there is the possibility I've missed a critical footnote].

Here's Japan's traffic per fixed broadband line since 2004:

Source: Japan Ministry of Internal Affairs and Communications 

A few observations:

• Bandwidth usage is growing, but not exponentially. The growth rate in the last year was 17%
• This is despite the fact that there was major adoption of FTTP in this period - as a share of fixed broadband it rose from 36% at the start of 2007 to 67% today (data here)
• Usage is low relative to the line capacity - 54 Kbps vs the average speed of 10.8 Mbps measured by Akamai, representing a utilisation of 0.5% (though certainly it will periodically spike far higher than this for any given line)
• It's also not the case that fibre delivered massive growth in per-line upload traffic - this peaked in late 2009, and has fallen almost 30% since then

It seems fair to call this a little underwhelming.  One factor may be that consumers are choosing to use the LTE mobile network rather than fixed connections. Indeed, it seems some consumers are seeing LTE as a complete substitute. According to Informa:

"NTT East and NTT West have been forced to slash their FTTH prices for new subscribers by an eye-watering 34% from ¥5,460 (US$66.70) to ¥3,600 per month to try and re-ignite their subscriber growth and stop the outflow of subscribers to cheaper LTE mobile broadband services."

Be that as it may, given that other countries are spending billions to replicate Japan's superfast fibre-optic infrastructure, presumably Japan is streets ahead of what's typical usage on copper networks? Another country that publishes usage stats is Australia, where the current government is investing massively to build FTTP (though 96% of households are still on cable or DSL).

Here's how Japan and Australia's per line traffic compares:

Source: Australia fixed lines and traffic from ABS. Japan as above

Note: Japan units converted from average kbps to monthly total GB

As of the end of last year, Australia's per-line fixed broadband traffic was roughly 70% higher than that in Japan. Despite a network that is undoubtedly technically inferior to Japan's, Australia is seeing robust growth. Of course all traffic is not created equal, but it seems fair to guess that Australia is getting more value out of the fixed internet than Japan is. By contrast, Japan's very expensive investment in FTTP doesn't seem to have delivered that much.

Network infrastructure is only as valuable as the usage it enables. Is it time for Japan to start envying Australia?

Netflix - bandwidth piglet

Netflix is notorious (at least amongst ISPs) as a bandwidth hog, consuming 1/3 of all North American internet backbone bandwidth. However, what's fascinating is that when you look at their usage from the perspective of an individual user's bandwidth, rather than the network as a whole, it's actually not overwhelming.

Some ISPs have an incentive to degrade Netflix traffic, since they have their own TV offerings they'd rather consumers bought instead. As a way to stop this happening, Netflix reports the performance it measures on different ISPs This data tells us an number of fascinating things.

1. Low bandwidth per stream
The bandwidth used per stream is low. In developed markets, the average stream is generally in the range or 2-2.5 Mbps. This isn't too much of a surprise, given that this is about typical for an SD TV picture. Whatever might fill up 100 Mbps bandwidth, it isn't likely to be standard TV

2. Performance driven by factors other than access speed
Most ISPs experienced a performance drop of approximately 5% in December (see Finland for example). Clearly this isn't because the access networks were downgraded. Rather it's likely because things got busy for the network (and Netflix) in that period, leading to congestion. This is important because it underlines that a poor experience for an end-user is often not caused by a problem in their internet access. There are many other parts of the system that can get congested

3. Access technology has relatively limited impact
While the numbers show a discernible difference between technologies, it isn't huge. In the US Verizon's FTTH-based FiOS service delivers Netflix streams at 2.10 Mbps. AT&T's U-Verse, based on FTTN, runs at 1.91 Mbps, not far behind. (Even the fabled Google fibre is only running at 3.35 Mbps for Netflix). Clearly the FTTH networks are capable of greater speed, and Netflix adapts to the available bandwidth, but in practice FiOS and U-Verse customers were having very similar experiences. This could be because there was a choke point somewhere else in the system, or the user had capped speeds (to husband data allowances) or the content was already playing at maximum resolution. No matter, the extra bandwidth was giving very limited benefit, even with a bandwidth intensive service like Netflix.

4. DSL compares well
Fourthly, the FiOS speed is all the more striking when compared to UK DSL providers. Of these only one (Everything Everywhere) was materially below FiOS' 2.10 Mbps. Of the other three, one exceeded it and two matched it (if we give the benefit of the doubt to Talk Talk, a whisker behind at 2.04 Mbps).

5 Stream speeds low relative to access network capabilities
These levels of usage (in the houses that actually use Netflix) are relatively low compared to the capability of the access network. Akamai measures average peak connection speeds at 29.6 Mbps for the US and 28.1 Mbps for the UK. While this isn't a strict apples-to-apples comparison to the Netflix figures, which are averages and will peak higher, it does suggest there's ample headroom even in today's networks.


Maybe massively higher resolution video and 3D TV will change this picture, and suddenly Netflix will be using much more? Perhaps. But even 3D TV will only require 12 Mbps, and this won't be widely used for a long time to come, not least because there just isn't that much 3D content yet.

Netflix may be a hog when it comes to the internet backbone. However, as we've seen, when it comes to the access network it's just a piglet.

EDIT:

The above post is as originally written, but MikeyB has pointed out something important I'd missed:

The Akamai average peak speeds are not averaged across lines, but but across IP addresses. Because IP addresses are dynamically allocated, one IP address might be applied to several lines in the course of a quarter (Akamai's measurement period), and Akamai's average-peak measure will only count the fastest of these lines. I'm not sure of the magnitude of this issue - it will depend on how frequently IP addresses are reallocated (particularly for fast lines), but there's no question that it means that Akamai's figures are a less good guide to the capacity of the average line than I thought, overstating it to some extent.

The other figure Akamai publishes is the average connection speed, which is undistorted by the 'peak' issues described above. However, it has its own issues. The average will be lowered by congestion at peering points, contention in the backhaul, server overload, interference with the end-user's wifi signal and various other things - none of these have anything to do with the speed of the access network, and thus this metric understates the capacity of the typical last mile connection.

Akamai's average speed will also be affected by simultaneous use in the home - if more than one person is sharing the connection, then the bandwidth 'perceived' by Akamai will also be reduced. Again, this will cause it to understate the speed of the access network, though may accurately reflect the speed available to a given person on a household.

Thus the capacity of the access network is likely lower than the peak speeds 29.6 and 28.1 Mbps for the US and UK respectively, and higher the average measured by Akamai, which is 7.2 and 6.3 Mbps.

Thus, I think the wider point that Netflix's bandwidth needs (at roughly 2 Mbps) are a long way below the capacity of the access network for the typical user stands. But my thanks to MikeyB for pointing out my error.

Fibre lobbyists think you're stupid

I've become acclimatised to some fairly blatant distortions of data and leaps of logic by some fibre lobbyists, but every now and then I'm still taken aback. Yes, I acknowledge it happens on both sides, but I would expect better of a company like Cisco.

I've been reading Cisco's 2012 report Get Up to Speed -How Developed Countries Can Benefit from Deploying Ultrafast Broadband Infrastructures, and came across this gem. The report offers examples of ultrafast broadband services in use today, including (p11):

"French company Erdenet supplies web-based courses that students can study at their own pace using interactive video and online collaboration tools. With an ordinary DSL connection, it is difficult to add rich media such as video, audio, and maps. Such applications require fiber connections."

Video, audio and maps require fiber?  Really?

For video, YouTube specify a minimum bandwidth of 500 Kbps. Of course higher speeds will give you better video resolution, but there's a big gap between 500 Kbps and the bandwidth of even a below average DSL connection. In the UK, less than 2% of ADSL households (roughly) are getting less than 7 Mbps in the evening busy hour (when speeds might be expected to be at their worst).

For audio, the picture is even starker. Even high quality audio streams typically use 320 Kbps or less.

The idea that maps require fibre I won't even dignify with a link. We've had maps online since the days of dial-up.

What's striking about Cisco's claim is that, while I've provided some technical data above, it's absolutely not necessary to know the claim is nonsense. Any DSL user reading that report will be well aware they're able to  use video, audio and maps.

I know Cisco aren't stupid - they can't possibly believe what they wrote. So it seems like the only other option is that they think we are.

'Compelling applications' for fibre - are we there yet?

Every year at about this time, the FTTH Council of Europe have their conference. They are the most enthusiastic proponents of fibre-to-the-home - they're the folks who claimed that it would be useful for remote surgery (see 0:50 of this). That's fibre to the home we're talking about ...

Anyway each year they present various assessments of the state of the European market. Part of the purpose of this (and it's a skilful bit of lobbying) is to create a sense of competition to see which country can rank highest in FTTH roll-out. The presentations have all sorts of rankings by different metrics, so almost every country can get some kind of recognition. It's like a school fête where almost every kid gets a prize. (The UK and Germany are the sulky kids who sit in the corner and won't even participate in the egg-and-spoon race).

The presentations also talk about some of the challenges that FTTH faces. Here's some disarmingly honest quotes from the Council (or their consultants) about FTTH from the last three years:

  2011: "No really compelling application that requires a fiber connection"
  2012: "No really compelling application that requires a fiber connection"
  2013: "No really compelling application yet"

Are you spotting the pattern here?

Now sometimes we're told this is a chicken-and-egg problem, that of course there aren't applications that need FTTH, because there isn't critical mass of FTTH to sustain them yet. However, according to the FTTH Council there are 107m FTTH connections around the world. Give or take, that's roughly the same as the number of basic broadband connections there were in 2003 (103m, according to the ITU).

By 2003 we had the iTunes store, Skype, streaming video, movie downloads, numerous IPTV services and many other broadband-dependent services that are critical features of the market to this day.

If 100m broadband households were more than enough to support the development of a plethora of compelling applications for basic broadband, why is it that 100m FTTH housholds don't seem to have driven any compelling applications at all?

Not only are we not there yet, it may just be the case that there's no 'there' there.

Does faster broadband encourage usage? Not so much

If faster broadband speeds enabled a cornucopia of marvellous services, we might expect countries with higher speed to spend more time online. However, the data from Europe doesn't show that picture:

Source: Akamai (Q3 2012) and Comscore (Dec 2012). Free registration required

As you can see, there's not much of a pattern (R² of 0.04, with a negative slope). In other words, there's no evidence here that, as between European countries, higher speeds are leading to greater use of the internet.

Of course one might argue that the ability to get things done quicker means that people with higher speeds spend less time online, but this certainly hasn't been the historic outcome - in the UK in 2006, broadband users were spending 60% more time online that dial-up users, for example.

I don't want to overstate my case here - there's lots of factors that drive hours spent online, and it may be that some combination of these represent so much 'noise' that they are masking a positive impact from higher broadband speeds. However, that in itself is interesting - it suggests that even if higher broadband is important, it is readily swamped by these other factors.

The above finding regarding time online is also consistent with evidence that broadband speed makes no detectable difference to the economic impact of the internet across a range of European countries. (See p4 of this)

The scale of the Facebook time sink

Nielsen has some figures on US usage of social media, looking in detail at July 2012. In that month, US consumers spent a little over 93bn minutes, or 1.5bn hours on Facebook, of which 1/3 was via a mobile device.

The US population aged 13 and over (Facebook's minimum) is roughly 252m, so this usage works out at 6.2 hours per month for every single American adult and teenager. The figure would of course be even higher for Facebook users in that group.

Another way to look at this is to compare to the working hours of the American labour force - 142m people in July. This includes part-timers and those working long hours, but to keep things simple, let's assume a 40 hour work week. This means the US economy has monthly working hours of 23bn, and monthly US Facebook usage is equivalent to 7% of this figure.

I am not suggesting that this 7% substitutes for working hours, though certainly some does. And some of it likely substitutes for study time. The sheer scale of Facebook usage means it must be substituting for a whole range of activities, be they productive or purely pleasurable.

Something to put on the balance sheet the next time you're being told about the (real) positive economic impacts of the internet.

What does the Center for Disease Control have to tell us about broadband?

It turns out that the US's Center for Disease Control publishes statistics about the portion of households that do not have fixed phones (I'll leave it to them to explain why).

The figures for June 2012 are now out, and they're quite striking. In the US 34% of adults live in a house with wireless phones only. Amongst those aged 18-34, the figure is around 54%. Those number cutting the telephony cord is also rising fast - around 4% per year.

While the speed of this transition may be surprising, the underlying reasons are not. The ease of use for outbound calls, large bundles of inclusive minutes and improving voice signal quality all mean that the mobile is a preferable way to make calls, and of course calling someone's mobile is the best way to reach them. Also, mobile phones are (in the developed world) increasingly seen as a necessity. The fixed phone has become an incremental, and optional expense - one that represents increasingly poor value for money as usage switches to mobile.

So what's this got to do with superfast? Actually, it's more of an issue for fixed broadband in general.

Firstly, the underlying costs of a fixed connection (either copper or fibre) are virtually the same, whether it's being used for voice, internet or both. Consequently, the margin on adding an extra service can be high - hence the popularity amongst operators of bundles. However, the telephony cord-cutting trend suggests that there's going to be diminishing contribution to that cost of a fixed connection from voice. Any Next Generation Access business plan would do well to be cautious about expecting voice revenues to make a meaningful contribution.

Secondly, the dynamics of fixed and mobile voice discussed above could also play out with fixed and mobile broadband. Mobile data usage is likely to become very widespread, and seen as necessary in just the way mobile voice is today. This sets up the decision to also have fixed home broadband as an incremental choice - an extra spend that must be justified by the extra benefits brought over and above just using wireless broadband at home.

The extra benefits are real - fixed broadband will bring greater speeds to most consumers, and for those using material amounts of data, the higher data allowances on fixed broadband will be valuable. However, for many consumers (perhaps those on somewhat tighter budgets), they may decide that the extra cost of these benefits is not worth paying. It is in this sense that mobile broadband is potentially a substitute for fixed broadband - not because it is technically as capable (it isn't), but because some consumers may choose to buy wireless broadband instead of fixed.