Newsletter: Get Real on LTE data rates

LTE data rates and Nielsen's Law. Is it holding up?

Prudent Advice: Get real on LTE bit rates

By Claus Hetting [claus@hettingconsulting.com]

An avalanche of LTE-marketing is in the making, and for those of us who have been in the mobile business for a while, it’s all decidedly déjà vu. Remember the UMTS marketers who promised us a whopping 2 mbps and the brave new world of the Mobile Information Society? I remember them well. In fact, I was one of them (working for Nokia Networks at the time).

Then reality intervened. In fact – as I see it – seven or eight years ago the entire mobile 3G (UMTS) industry was inches from disaster: Massive spending on licenses and a litany of technical problems. 3G nearly went to wrack and ruin before it had even begun.

Enter stage right HSPA, and for the first time, 3G operators have something that people actually want to spend their extra, hard-earned, financial-crisis-ridden money on. And it’s not voice over 3G (what’s the difference to 2G anyway?). It’s real mobile broadband at respectable bit rates.

Marketing the wrong speeds

The next big push is towards LTE. Lots of mobile operators in Europe and the US have committed to rolling out LTE, and some are building commercial networks already now - here is an example from Scandinavia: http://www.telecoms.com/10423/tele2-telenor-to-build-swedish-lte-network. Although we call it 4G, it is not really – it’s an evolution of 3G (to be honest I’m still not really sure why, but it is still strictly speaking incorrect to refer to LTE as 4G, so I will try not to do that).

Anyway, here’s the rub: The mobile industry is marketing LTE as being capable of 100 megabits per second downlink speed at least. For uplink most will quote 50 megabits. Either is – in my view – a gross overestimation (even though it is true that the standard speaks of such rates and even higher ones as its so-called target rates). Déjà vu? The entire industry is now likely to see history repeat itself and broadcast another set of hyped expectations.

And what is more, I personally believe that some decision makers are at risk of being lured into business ventures that promise much more than they deliver. CEOs of the mobile industry, beware! The business case for millions of subscribers at 100 megabits per second is a lot different that thousands of subs at a fraction of that speed.

Now there are lots of articles and news items out there about high LTE data rates – here’s a particularly bad one, which calls itself “the simple truth about LTE” and still quotes very high rates: http://unite.nokiasiemensnetworks.com/simpletruthLTE. This is from a vendor, but in my view it is typical of the types of messages that are being pushed.

Real rates & real information

So what bit rates can we expect from LTE? To get the real picture you need to do some digging into reputable sources - and that means proper technical literature. I have quoted a few sources below and in my view they are all trustworthy and – most likely – accurate. They are all readily available, although some of them cost money.

The first comprehensive study of LTE bit rates comes from the vendor community itself. It can be found here: http://www.lstiforum.org/media/event.html?listNum=176. LSTI is the industry organisation responsible for testing LTE systems, and perhaps as close as we get to real life LTE performance test results. It is good and it is free.

Flip to slide 15 of the presentation (not that the rest is uninteresting, but anyway) and you will find the answer: The real downlink rates (for LTE FDD) – as estimated by the LSTI – range from 25 megabits to less than 5 megabits per second depending on where you are in the cell and how many users are active. It appears that these estimations apply for 10 MHz in bandwidth & 2x2 MIMO - but this is not totally clear from the presentation.

Hard facts in hardcover

Another useful source is the hardcover textbook bible “Beyond 3G: Bringing Networks, Terminals, and the Web together” by Martin Sauter and published by Wiley. Mr Sauter is a well known mobile technology expert. In case you are curious can find his blog here: http://mobilesociety.typepad.com. Unfortunately, to get to the truth about LTE, you have to first buy the book, e.g. at Amazon here, and it is not cheap. But it is probably worth the money.

Flip to page 121 of the book and you will find a table listing most mobile data technologies since GPRS and their bit rates. The table cites a user data rate range from 3.2 to 10 megabits per second for LTE at 10 MHz bandwidth and 2x2 MIMO depending on cell load. In the case of LTE at 20 MHz and 2x2 MIMO the range is from 6 to 20 megabits per second.

And there are other sources citing similar numbers. To cut to the chase: A good rule of thumb that I use – which is by the way also supported by my very smart colleagues & friends at www.rewheel.fi - is that LTE will provide data rates that about are 3-4 times higher than current HSPA rates. Also do not forget that real, average HSPA data rates are not the 7 or 14 mbps claimed by marketers but a much more modest 2 mbps (among others the Swedes are particularly good at monitoring real mobile broadband data rates. After thousands of measurements for each service provider nobody seems to offer anything better than about 2.5 mbps on average despite claiming 7 mbps or more.If you're interested in this I can provide a reference. Moral of the story? In the Internet-connected world is getting very, very difficult to get away with marketing nonsense. Sooner or later, the truth will out.)

I guess I don’t have to point out that all the real numbers are a far cry from the 100 megabits that many have started to brag about. My advice to the industry is simply this: tone down the messages and get real on the data rates!

Complications, any one?

As an engineer with 15 years of experience in mobile radio, I am pretty confident that these numbers are roughly right. Most of the figures come from a mix of simulations and calculations, probably with a measure of engineering common sense thrown in. LTE will undoubtedly be a leap forward but it is not the Holy Grail of mobile broadband and is not likely – at least in the short term – to break any light speed 100 mbps barriers except in the lab.

Another contentious LTE issue – one of many - is the applicability of MIMO. Yes, MIMO is a sophisticated evolutionary next step for mobile networks. But the efficiency of MIMO may not be as high as many people expect, because MIMO – in order to work – requires that sufficiently different signals be received at the multiple antenna locations.

This may not be too difficult on the base station end, although more antennas will be needed [this alone is not uncomplicated and not all existing radio sites may be suited for this]. But MIMO user terminals will also need to have multiple antennas that are somehow physically separated by enough space to receive uncorrelated signals. This is not necessarily simple or even always possible.

Comment May 2011: Since I wrote this piece in 2009 I've been doing a lot more research on LTE and TD-LTE and more speed test results have become available. According to my most recent research and information it is likely that LTE (and TD-LTE) services may reach average speeds of some 30-35 mbps for medium loaded networks. I intend to publish a proper research note on the issue - as a Mobile MicroReport - as soon as I get time to write it!

Nielsen’s Law: Today’s market wants Internet access at 35 mbps

By Claus Hetting

The true demand for broadband Internet access in developed markets already exceeds 35 mbps per household. The number is derived from the so-called Nielsen’s Law and supported by data from e.g. Sweden, where fibre-based broadband is already widespread, says Ralf Pütz, fibre technology expert from the Sacoin consultancy in Germany.

Internet users in developed markets already want data rates that are well into the double digits. This is the opinion of a leading German fibre-networking expert, Mr Ralf Pütz of Sacoin. Mr Pütz was speaking at the Breitbandstage broadband conference in Hamburg, Germany last month.

- The number works like this: The market today wants speeds of around 35 mbps. If you for example provide a household with 100 mbps of access, the users will typically consume 35 mbps. Conversely, if you are not capable of delivering 35 mbps, you’re not cashing in on the full potential of the user, said Mr Pütz.

Sacoin is one of a few companies providing high-speed fibre networking (FTTH) solutions for underserviced communities in Germany. Many German villages and rural areas are today served only by DSL at a comparatively low data rate of 360 kbps or less.

Nielsen’s Law holds

Mr Pütz says that the 35 mbps value is derived from the so-called Nielsen’s Law. The American Ph.D. and IT guru Jakob Nielsen penned Nielsen’s Law in 1998. The law – which is the broadband equivalent of the better-known Moore’s Law of computing - says that the demand for bandwidth will grow by a rate of 50% per year. The law only deals with Internet access, and does not include other IP-traffic such as for example digital TV over IP.

Nielsen derived his law in 1998 after mapping out the development of Internet access speeds for 15 years. Points on the curve starts at a diminutive 300 bps in 1983 and ends at 16 mbps in 2008. Nielsen’s original curve – which can be found here - very closely follows the 50% per year growth rate over a period of 25 years. If extrapolated to end of 2009 the curve reads approximately 35 mbps. Mr Nielsen does not provide an explanation for why bandwidth demand seems to grow at this rate.

In February 2008 the FTTH Council [link] – the international industry organisation for FTTH equipment vendors – commissioned a study of Nielsen’s Law, and the results have provided some empirical proof that the law appears to hold. Several countries – including Sweden and France – have seen annual growth rates of 50% or more. The full report as provided by the telecoms consultants of Ventura Team can be found here.

Video is driving the demand

According to the report Swedish households with fibre-based broadband consume three to four times as much bandwidth for surfing and entertainment compared to e.g. households using DSL-based connections only. The bandwidth is used especially for Internet-based video and TV as well as superfast browsing.

- The results clearly show that when you give users access to more bandwidth then they will for sure use it up to 35 mbps or more. That is why access to broadband at very high speeds is really important, says Mr Pütz.

Date: 09-11-2009

Author: Claus Hetting

19-12-2011

HETTING selected for 3G project in Asia

HETTING to provide market-leading mobile technology consulting services

06-12-2011

HETTING selected for project in Russia

International Finance Corporation (IFC) selects HETTING for due diligence services

01-12-2011