Author: Martin

Compared to other countries in Europe, prices for mobile data services in Germany are still pretty high. A gigabyte of mobile data can be had for around 10 euros if you look around a bit (but still quite a difference for the unlimited amount of data you get for 18 Euros in Austria, for example). Anyway, how long does it take to use up this gigabyte?

In Germany, LTE is available at speeds up to 300 Mbit/s now when allowed by the contract and when being close to the center of a cell that has carrier aggregation with 2×20 MHz activated. If all those conditions are met, 1 GB of data can be downloaded in  in less than 30 seconds. But let's make a more realistic example, let's say I get 50 Mbit/s out of the link, which is not so unusual in many places. At that speed it takes around 3 minutes to use up my data bucket for the whole month if I really wanted to.

Now let's do an interesting thought experiment with that number: How many users could do that in one month with one base station site? A month with 30 days has 43.200 minutes. Divided by 3 gives you 14.400 users who could run that exercise sequentially. On the other hand, most cells have 3 sectors, each with an independent data stream so if those users are neatly distributed around the base station site that number goes up again. And now multiply that number by the average 1 GB data bucket price.

Of course networks are not fully loaded and due to interference from neighboring cells and distance from the base station not all users get 50 Mbit/s out of an unloaded cell. But even if the number is an order of a magnitude lower it is still impressive.

A couple of days ago a reader left a comment to a previous post on LTE 900 MHz in the Netherlands that an LTE network now also went live in the 450 MHz range in Finland. Thanks for that! I was intrigued so I had a closer look. The network seems to address mostly businesses and moving vehicles at first and private summer house owners in Finland a bit later.

What's The Deal For Today's Users?

I tried to find out which band is used for that network but couldn't find a definite source. So I assume it's FDD band 31. If that is indeed the case then it's at 5 MHz carrier at most (for which I don't have a confirmation either so it's a guess, too), compared to 10 MHz used in the 800 MHz band and 15 to 20 MHz carriers used in band 3. In other words capacity and top speeds are quite limited. So from that point of view nothing to write home about. On the other hand a 5 MHz LTE carrier in a summer house far off the beaten path is better than nothing at all, at least today. With rising bandwidth demands, however, a 5 MHz carrier won't cut it for long even in far away places…

The historical perspective

Oh yes, why only 5 MHz? It looks like this is a spot left over from old analog networks (e.g. the C-Netz in Germany, which used 451,30–455,74 MHz and 461,30–465,74 MHz according to Wikipedia. This range is not fully overlapping with LTE FDD band 31 but that's not very surprising because in the 1980's, there was no common European, let alone, worldwide standard. But from a historical perspective the 5 MHz bandwidth is interestig because I always imagined that more bandwidth was used for these networks. And then GSM comes along for which 25 MHz of bandwidth was reserved for each direction, i.e. 5 times as much as for the networks that were in use at that time. Quite a bold move for the time I would say.

Half a decade ago, network operators in Europe have started to use UMTS not only in the 2100 MHz band but also in the 900 MHz band that was, up until then, only used for GSM services. A good example is France with rural coverage and big cities in the UK. Now it seem seems LTE is also getting some traction in what is called LTE band 8. According to a report by Telegeography, T-Mobile Netherlands has started using LTE in the 900 MHz band in Amsterdam. An interesting move, just makes me wonder how much spectrum they have available. On the device side LTE band 8 is already supported by a number of devices, for example by the current generation of Apple devices. In other words there will be immediate use of the extra resources.

When I recently had problems keeping a UDP session alive over a NAT router I found out that my Raspberry Pi acting as the NAT router only keeps port mapping for the UDP session for 3 minutes. If there is no traffic during that time the mapping is discarded and all incoming packets afterward will no longer be delivered. While that's probably o.k. for most applications it didn't work for me. Fortunately it can be easily changed.

The following commands shows the current UDP session timers variables and their values:

sudo ls -l /proc/sys/net/ipv4/netfilter/ip_conntrack_udp_*

sudo cat /proc/sys/net/ipv4/netfilter/ip_conntrack_udp_*

On my Rasperry Pi I get 30 seconds for the initial timeout and and 180 seconds once the stream is established. The following command changes the timeout for new streams to 20 minutes:

sudo sysctl -w net.ipv4.netfilter.ip_conntrack_udp_timeout_stream=1200

And to find out which TCP and UDP streams are currently in the NAT translation table the following command can be used:

sudo cat  /proc/net/ip_conntrack

The second value in each line is the number of seconds remaining before the entry is removed if no further packets arrive that are part of the stream.

And for more details have a look here.

In the past couple of days I've had two posts on how we are now past the human subscription peak in Germany and also past peak telephony, both based on numbers of the German telecoms regulator (RegTP) report for 2014. In this third part let's have a look at some other numbers and how they compare to those of previous years that I highlighted last year in this post.

First up is the rise in data volume: In mobile networks in Germany, the total data volume has risen to 393 million GB in 2014, up from 267 million GB in 2013. That is an increase of 47% compared to increases of 70%, 56% and 53% the years before. In this context it is interesting to note that the number of 3G/LTE capable devices and subscriptions has risen by 32% last year from 39.6 million to 52.6 million. This makes me wonder how much of the 47% increase in data volume is due to growing use per subscriber vs. growth due to more and more people opting for a data subscription. I would say a fair amount of the increase goes to more people using data services. Even non tech-savy and older people are now crazy for messaging services like Whatsapp and kids getting their first smartphones don't even bother with voice and SMS to begin with.

On the infrastructure side there are 28.700 LTE base stations in Germany now, an addition of 10.000 base stations in just one year. 80% of the population now have access to the LTE network of the market leader. Overall, 92% of the population are now covered by at least one LTE network and 74% of the population can get wireless services with speeds of at least 6 Mbit/s.

On the fixed line side, data use has grown by 22% from 22 GB per month per line to 27 GB per month. As last year, that means that 30 times more data volume is consumed per subscription on the fixed line side compared to on the wireless side. As there is usually one fixed line subscription per household but several mobile subscriptions the difference on a person by person basis is less but still significant.

On the spending side, 7.4 billion euros were spent in Germany in 2014 on network infrastructure, up from 6.3 billion euros in 2013. While that sounds like a massive increase it's worth to point out that in 2007 and 2008, before LTE was deployed, spending was at a similar level (7.1 and 7.2 billion respectively). and

On the employment side, the number of people working in the sector decreased by 1800 employees to 168.900 in Germany. Over 10 years the overall reduction amounted to 56.0000 employees. Quite a hefty number.

Let's close this post with two positive numbers: In some rural areas, fixed and wireless broadband networks are still nowhere to be seen and thus there are 30.000 subscribers in Germany for Internet over satellite. And on the network modernization side, 41% of fixed line calls are VoIP end to end now.

A quick not on the way wireless changes. Back in November 2012 I had a post on Wi-Fi tethering replacing my 3G stick on the train. Now my connection manager software tells me that I haven't looked back since then as it says I last used cellular connectivity over USB 2 years ago (see screenshot on the left). Yes, Wi-Fi tethering to a smartphone has become just so easy and convenient to use. And by the way, most of the route to and from work is now covered by LTE so UMTS wouldn't cut it anyway anymore. It's hard to believe I already wrote the post linked above over two years ago, time passes so quickly…

After acquiring the E-Plus mobile network in Germany, Telefonica O2 has started integrating the two physical networks. For their customers the first visible change is that Telefonica has enabled national roaming between the two networks. There are several ways to do this so I had a closer look which option Telefonica has taken in Germany.

As reported by the press, national roaming has been enabled for the two formerly separate GSM and UMTS networks but not yet for LTE. In Cologne, there's advertising in the streets now to make customers aware of it. And indeed the two UMTS networks are now open for national roaming while the LTE networks are still separate and registration attempts with a SIM card of E-Plus are rejected on a Telefonica LTE cell. At this point in time in Cologne, however, the GSM networks are not yet shared and my E-Plus SIM card is rejected in Telefonica's GSM network. From a policy point of view that's quite interesting and shows that mobile Internet access is now in the focus rather than 2G voice telephony.

To make devices aware that they can roam between the networks, there's the Equivalent PLMN (Public Land Mobile Network) ePLMN mechanism defined in 3GPP. The ePLMN indication can be put into the GSM, UMTS and LTE broadcast information which is what is done in Austria for example to indicate national roaming between ex. Orange, 3AT and T-Mobile. Telefonica in Germany, however, has chosen not to go down this route as I could observe no ePLMN information in the broadcast information of any network technology.

Instead, ePLMN information is sent to the device at the end of the registration process in the Location Area Update Accept and Routing Area Update Accept messages. I played around a bit and this works just as well. As designed, my mobile went from an E-Plus cell to an O2 cell despite manual network selection to E-Plus.

While being 3GPP standards compliant not all customers are happy with this in practice. As reported in this article, some Telefonica/O2 customers are not happy to find their mobile on the E-Plus side of the combined network which at times and in some locations seems to be more congested than the O2 network that is also present. But once in the E-Plus UMTS network the only way to automatically go back to the O2 UMTS network is to loose E-Plus UMTS and GSM network coverage. It would of course also be possible to define neighbor cell relationships between the two networks so the mobile could find cells of the 'other side' but I guess that's too much trouble for an interim solution.

Network sharing is also bad news for subscribers in dense population areas where only one of the two UMTS networks has been available so far. Due to the sharing the network now has to cope with twice the numbers of subscribers. According to the article above that has some rather unpleasant slow-down effects at times.

A quirk caused by sharing the two UMTS networks but not the LTE networks is that once a device is in the 'partner' UMTS network it is unable to go back to the LTE network of its home network operator until coverage of the shared UMTS/GSM network has been lost. As the LTE networks are not shared it would not even be possible to announce neighbor cell relations ships. I can see why that will make some customers with patchy LTE coverage quite unhappy.

So while national roaming is a good idea to extend coverage for customers and probably a requirement to shutdown the E-Plus network over time it will be interesting to observe how many customers are affected by the negative consequences of a move that the marketing departments try to give a positive spin.

As reported in the previous post, another spectrum auction has started in Germany this week, this time with only 3 companies being allowed to bid for the spectrum. Two things make this spectrum auction especially interesting for me. First, all of the GSM 900 spectrum and quite a bit of the GSM 1800 spectrum is re-auctioned as the licenses awarded a decade or two ago are expiring. So it's going to be interesting to see who wants to acquire how much spectrum in the pretty narrow GSM 900 band that is not very suited for broadband Internet services because it will have to be used for continuing the GSM narrow-band service there for the foreseeable future. The 1800 MHz band is a different beast as it's broad enough for high speed Internet services of several network operators and already used in Europe for that purpose in addition to GSM.

The second interesting thing for me in this auction is that the German Regulator (BNetzA) requires each company that acquires spectrum in the new 700 MHz (digital dividend 2) band to cover 98% of the population with their mobile broadband Internet service with a speed of at least 10 MBit/s per customer on average, and 50 Mbit/s per sector of a base station. The later requirement means that at least 10 MHz of spectrum has to be used per sector. For the details have a look at the 250+ page rules and requirements document for the auction.

Today, we are still quite a bit away from that goal. According to the regulator's report for 2014 that was published a couple of days ago, 92% of the population is now covered by at least one LTE network and the market leader's LTE network covers 80% of the population. According to the auction rules, EACH network has to cover 98% of the population in three years time, however, so from that point of view there's still some work to be done.

And finally I also find it quite interesting that the rules also go into the details of what statistics each network operator has to annually deliver to the regulator, including the requirements to supply SIM cards and methods for the regulator to make their own assessment how well each network is deployed.

Yesterday, the 2015 spectrum auction for wireless network operators has started in Germany. In addition to the re-auctioning of spectrum in the 900 and 1800 MHz band due to decade old licenses expiring, new spectrum is auctioned in the 700 MHz band (Digital Dividend 2) and some extra uni-directional spectrum in the 1500 MHz band. Hopefully, having only 3 companies bidding for the spectrum will not drive up the auction results to unreasonable levels as in the past. Anyway, Teltarif has published a good report about the results of the first day here and they'll probably follow the proceedings and comment on a regular basis. Worth watching in case you are interested in spectrum auctions. Their posts are in German but Google can help with the translation… And in case you are wondering about the T&C's of the auction, they can be found here, again, unfortunately in German only.

A couple of days ago I started my analysis of this year's report of the German telecoms regulator (RegTP) with a first post on that we are past "peak telephony". The report also clearly lays out that we are past the "human subscription peak" as well.

10 years ago in 2004 there were 89 million mobile subscriptions in Germany. The peak was seen back in 2011 with 142 million subscriptions. Since then the number of subscriptions have gone up and down a few millions year on year and in 2014, 139 million subscriptions were counted. In other words there is no growth anymore despite the push for mobile devices in addition to smartphones such as a tablets that also have cellular connectivity.

So should we see growth in this area in the future again it will probably come from other areas. Machine to machine for example. In other words the number of SIM cards might from now on be a good indicator of how much and how fast non-human machine communication gains traction.