Your browser is not supported. Please update it.

4 December 2014

Vodafone Liberty – The architecture behind a modern marriage

The news that breaking this week about Liberty Global talking to Vodafone about a possible sale, is something that has been on our minds since 2008, the year Faultline first pointed out that Vodafone needed more fixed lines in Europe. (See separate story in today’s issue).

At the time we considered that Vodafone could buy the cable companies in countries where Liberty Global had no presence or the remainder where they already did have a presence, and that it could partner with Liberty Global for an MVNO presence, so that the two could jointly put a squeeze on incumbent telcos. You can say we were not being ruthless enough to envision a merger as something which was quite inevitable once you have gone through that logic.

So instead we forecast and cajoled Vodafone to buy Kabel Deutschland and lo and behold, it went and did that five years later. Since then it has picked up Ono and built out fiber in Spain, partnering Orange in the process. Vodafone also plans to build fiber in the Netherlands and as we have said all along, should have acquired BT in the UK. But why?

The answer is in the physics and in everything that we write about at Faultline and this is the “big” deal, the deal that decides on the fate of European communications, just as the Comcast Time Warner Cable deal decides that fate of the US. The two deals are inextricably linked, and the regulators will both be sweating on both sides of the Atlantic over the decision, eager to copy ideas from one another.

First the rather imperfect basic physics. Cable carries a lot of broadband capacity, which for the most part is not reliant on fiber being quite so close to the home, as it is for say Telco delivered broadband. A typical cable offers almost 1 GHz of spectrum in a protected cable which doesn’t leak or interfere with other nearby cables. If you have a spectral efficiency of 6, which is quite possible on a DOCSIS modem, then that’s 6 Gbps right there. However the specification for DOCSIS 3.1 anticipates capacities of at least 10 Gbps, with 1 Gbit/s upstream, using 4096 QAM modulation and new channel sizes.

Two of these can run side by side and not interfere with one another and once you have ripped up the road to put it in, the price to lay a second cable alongside it is a fraction of the first one, due to using cable conduits that encourage multiple parallel cables, often as many as a dozen or 16 cables can be accommodated in a single conduit. So you can potentially lay a further 160 Gbps of capacity through a single hole in the ground, and this can reach, depending upon the quality of cable that you use, for about 800 feet, without repeaters, to QAM, CMTS or CCAP. And then you attach it to fiber, and backhaul the signals back to the regional headend.

The only problem with this architecture is that these cables are then used for multiple homes, but this lets an operator carry out constant changes of loop length inside a cable environment, so that its capex is in line with the rise in customer usage.

We all know that Telcos have had trouble keeping up. They have a star configuration, from a single backhaul point, and unprotected twisted pair which interfere with each other’s signals when they are placed in a bundle. This architecture seems to top out at 1.0 Gbps over a distance of 70 to 100 feet. That means that fiber has to come a lot closer to each home and makes the cost of build out fantastically high, when compared to cable – Telcos the world over have appeared to be reluctant to make the requisite investment to overtake cable, except where there are lots of greenfield sites, which can have fiber direct to the home or where labor for digging up roads is very cheap.

It is these physics that dictate the economics. Wireless signals have come a long way and have been even more heavily engineered, but where we are at right now is that a single radio in a base station can carry 300 Mbps.

This is only possible in LTE using 4 x 4 MIMO, and when it can achieve 64 QAM signaling, on the occasions that it does not have to fall back to less efficient signaling to make the signal more robust. This means that at a high spectral efficiency, a cellular system can operate at about 400 Mbps, using about 25% of this on control and signaling housekeeping, leaving about 300 Mbps usable in a 20 MHz transmission. This is more than twice as efficient as cable, because it is so well engineered, but it will not get any better as it is already approaching Shannon’s law. This speed is now fixed.

But it can only do this in a single chain, in other words to one device at a time whether the signal is FDD or TDD (FDD is offering some of the capacity to each device simultaneously, TDD is offering all the capacity to each device in sequence – but it’s the same capacity). Using carrier aggregation, this can go up because essentially you are sending more signals, adding some on perhaps an extra 10 MHz in another spectrum slice, using a second radio. This is why we have just heard that Qualcomm has a 450 Mbps modem chip.

But if we take a country which has 70 million handsets and 54,000 Macro base stations (for instance), this means there are almost 1,300 handsets per base station. Most base station have 3 radios. If the peak phone use is when 25% of people are using their phone (for voice or data) at the same time, the average base station may have to cope with over 300 users simultaneously, or 100 per radio, suggesting they could only each get 3 Mbps each. It’s a huge disparity with fixed lines, and is the reason that cellcos are now wishing they owned fixed lines.

Currently when people use their phones at work and at home they are offloading about 70% of that data capacity to a WiFi network. So if you relieve the network in that way, the performance per customer gets closer to 10 Mbps at peak. And if you build out another 100,000 small cells, as some operators like Vodafone, say they are doing, that would give each consumer in peak hours closer to 30 Mbps, and although mobile companies do their best to fudge what you are actually getting, this would be heaven compared to what consumers are getting today. Just a few years ago it was closer to 1.5 Mbps in real terms.

Vodafone said in its results last week that it was close to achieving its target of installing 70,000 new small base stations under its Project Spring initiative, but among its 433 million customers, that amounts to a tiny improvement in performance at peak times.

When  you compare these cellular speeds with the speeds that are needed to play video, then there is a major problem, because a lot more than 25% of consumers watch video at the same time. We all know that Netflix and Hulu combined, plus YouTube, makes up most of the US internet in peak TV viewing hours. No cellular network that is working today could withstand that kind of usage or get close. We talk about this as the video Tsunami that is approaching cellular networks.

In comparison each cable home is getting a share of 10 Gbps, perhaps 300 Mbps now and 1.5 Gbps when DOCSIS 3.1 is widely deployed. Cellular will not hit this level until multiple tens of millions of small cells have been deployed and backhauled – at the rate that Vodafone is going, despite the fact that its famous Project Spring is being funded by the massive windfall profits it achieved at Verizon, it will still take between ten and 20 years and be late into the LTE Advanced deployment or even beyond.

Vodafone now has 11.2 million broadband lines in the world with some 10.5 million of those being around its core markets in Europe. By buying Liberty Global, at a stroke it goes from potentially using the same backhaul points which support those 10.5 million lines in Europe, to having the potential at every TV home in cable, right now that is 15 million broadband subs, but which could easily extend to 100% penetration in its 21 million pay TV homes. In other words it jumps from 10 million potential backhaul points to 30 million in Europe, crashing its infrastructure costs considerably and essentially tying up so many broadband homes that no other operator in Europe can keep up, about 14% of all European households, and perhaps 17.5% of all European pay TV homes (the richer homes).

And remember that Vodafone only acquired 6 million of that 10.5 million in the past 14 months with the acquisition of Kabel Deutschland in Germany and Spain’s Ono. It bought the others years ago when it acquired Mannesmann in Germany then more recently Tele2 broadband lines in Italy and Spain. It has also been building out fiber in Spain, Portugal and the Netherlands.

So in a little over a year, IF it buys Liberty Global, Vodafone will have gone from 4.5 million fixed lines, to certainly over 25 million, with the potential, as we say, to reach over 30 million. It would not only be the largest cellular operator in Europe, but the largest fixed line operator as well, dwarfing Deutsche Telekom, which has just 18.8 million broadband lines across all of Europe, and leaving every other player in their wake.