Digital Dividend for Rural Areas First Says German Regulator

Recently, German legislature has approved the reassignment of terrestrial television frequencies to wireless Internet services (the digital dividend). Heise news now reports that the German telecoms regulator (the Bundesnetzagentur) has started speculating about potential obligations for this frequency range.

As the 800 MHz band is useful in both rural and urban areas, the regulator wants to ensure that it is not only used for cheaply covering large cities while rural areas are forgotten. Therefore, they are thinking about mandating that the band may only be used in urban areas once 80% of rural cities with less than 5000 people or currently uncovered zones defined by the government are covered with a minimal speed of 1 MBit/s (user peak speed?). An interesting move to ensure the intent under which the frequencies were freed up will be met.

Now let's see how quickly the frequencies will be auctioned.

Base Station Numbers

Here are some interesting stats from Vodafone they have published earlier this year on the number of base stations they have in Germany. In the press release they mention that they have more than 20.000 GSM base station covering 99% of the population and 13.000 UMTS base stations covering 80% of the population.

Taking into account their current customer base (or more precise the number of SIM ards) of around 34 million let’s do some maths with the numbers:

For GSM: 34 million customers divided by 20.000 base stations means that each base station on average serves around 1700 customers. Most base stations have three sectors, so each sector covers on average 1700 / 3 = 566 customers.

Number of voice minutes on average per user per month: 111 minutes. So each sector of a base station would on average serve 111 minutes per month * 566 customers = 62900 voice minutes. To count in effects such as busy hour, let’s say those minutes are handled in 15 hours (a gross simplification…) of the day. 62900 / 30 days / 15 hours = 139 voice minutes per sector per hour. Given that most sectors are equipped with 2-3 Transceivers which can each serve between 6 and 8 voice calls would result in more than enough average capacity and still leave ample room for GPRS/EDGE services. Actually, I think the number looks rather low compared to what should be possible. Not sure why that is…

Note that in big cities, the base stations are probably much higher loaded due to the higher population density. That’s easily missed when working with averages.

Concerning the UMTS numbers, it’s a bit difficult to make meaningful similar calculations here. Vodafone says they have 5+ million UMTS customers. That probably means they have that many customers with a 3G phone. However, it says nothing about how many people have locked their phone to 2G for various reasons and also gives no indication on how many people use data intensive 3G Internet access. So I'll leave those numbers as they are for the moment.

Frequency Bands and Bandwidths

I’ve asked myself today of how much space there really is in all the different bands being discussed for mobile services in the future compared to the bands used today. Here’s a list of some of them:

Digital Dividend Band in Europe:

790 – 862 MHz (with a rumored duplex gap of 12 MHz)
30 MHz for each direction

Classic 900 MHz band in Europe for GSM
880-915 uplink, 925-960 MHz downlink
35 MHz for each direction

Classic 1800 MHz band in Europe for GSM
1710-1785 (uplink) 1805-1880 (downlink)
75 MHz for each direction

Classic 2100 MHz band in Europe for UMTS

1920-1980 MHz (uplink), 2110-2170 (downlink)
60 MHz for each direction

ITU-2000 band for LTE
2500-2570 (uplink), 2620-2690 (downlink)
70 MHz for each direction

So when comparing these numbers, the European digital dividend band looks pretty small with its 30 MHz. That’s 10 MHz for 3 operators or one operator with 20 MHz and another one with 10. So it won’t even be enough for two network operators wanting to use 20 MHz LTE carriers. Further note: Neither the 3GPP UTRAN spec (TS 25.101) nor the 3GPP LTE spec (TS 36.101) do yet contain an operating band number for this region.

In the US, the digital dividend band seems to be even narrower. Both Verizon and AT&T only have 2×10 MHz chunks. Not a lot of room to maneuver in.

Operator QoS for Skype & Co.

Recently, Nokia has announced that they will integrate Skype into the Nokia N97. Reactions, obviously, have been mixed. But I think the trend is difficult to stop, if not on this device it will be on another or in another way entirely. Some network operators have responded by announcing that they are thinking about introducing special tariffs which would include VoIP. But there is one thing over the top VoIP (i.e. non-operator circuit switched voice) doesn't have today, and that is the possibility to ensure the quality of service (i.e. latency, delay and jitter) especially over the air interface.

However, with a bit of imagination it wouldn't be too difficult to set this up. Here's one example of how it could work: In tariffs that take VoIP into account, the network could establish a secondary PDP context (UMTS) or a dedicated bearer (LTE) when it detects IP traffic of VoIP applications. This prioritizes the voice IP packets over other IP packets in the data stream of the user and also over IP packets of other users. Most mobile network operators already have deep packet inspection devices in their networks for all sorts of things and these could easily do the job.

I think it's an interesting technical possibility, let's see if somebody picks it up and puts it into commercial reality.

Finding Dormant Cells

Having stayed a bit outside of Barcelona for a couple of days before the Mobile World Congress back in February, I ran across a phenomenon again that I call the "Dormant Cell" issue. While a "dormant" cell seems to be up and running for both the mobile devices and the RAN control center, it is somehow locked up internally and is not functioning properly. In my case outside Barcelona it was a UMTS cell and the signalling to establish an Internet connection worked just fine. However, subsequent data transfers did not work at all. A kilometer or so away things worked again as the mobile selected a different base station.

One would think operators would detect such cells quite quickly and reset them. However, it seems that in practice cells sometimes do not send alarms to the operation center so the personnel is oblivious to the issue. In my case the situation remained that way over three days. Now try to report such a behavior to the operator's first line support staff…

To catch such issues, some operators run statistical counter analysis like for example to compare daily data transmission volume per cell. If suddenly a cell shows abnormally low values the analysis program generates an alarm so the network operation center can take a closer look. Unfortunately, not everyone seems to do that or at least not very often, like the operator of the network I used in Spain.

Agreed, it's quite a number crunching effort when you have several thousands of cells in your network and requires a big database to compare current counter values against those measured in the past. But benefits of such analysis go far beyond just finding dormant cells. This way, it's also possible, to give just one example, to follow rising use of certain cells in your network and to predict when it will run out of capacity. That way the network can be upgraded before the cell runs into saturation. The principle does not only apply for Internet access but also for voice calls.

I picked the Barcelona incident to start the post but quite frankly, it is not the first time I've seen such behavior and not only in a single country. Looks like this is a general phenomenon experienced not only with a single RAN vendor.

Vodafone and Telefonica – Infrastructure Sharing Light

Vodafone and Telefonica have recently announced a network sharing agreement that has gone through the press with lots said about it, but few have actually taken a closer look what kind of network sharing they have in mind. Fortunately, the Vodafone press announcement contains quite a number of details about what exactly is shared.

Unlike other infrastructure sharing deals announced by others in the past where everything is shared from mast to base station (and which I am quite critical of), this deal foresees to mainly share masts, power supplies and shelters. In many places this has already been done for years and this agreement looks like it will mainly tie the two companies closer together to lower costs while at the same time it ensures their independence on how much capacity they want to have at each site.

An interesting side note in the announcement is that the companies will also look into the possibility of sharing backhaul. While not yet part of the deal, I think this is a good idea, as it doesn't make a lot of sense to dig up the street several times to lay in fiber. I wonder if that's o.k. by the regulators.

And finally, I wonder what happend to a very similar deal between Vodafone and Orange which was announced just a year ago!?

HSPA Uplink Speeds from 1 to 11.5 MBit/s

A couple of days ago I had a post on the currently defined downlink speeds of HSPA from 1 to 80 MBit/s. As the uplink is just as important as the downlink for many application, 3GPP has also kept improving data rates in this direction. The following table shows the speeds defined up to Release 8 of 25.306.

  • Category 1 , 10ms:  0.7 MBit/s
  • Category 2,    2ms:  1.4 MBit/s
  • Category 3, same as 2 but only 10ms TTI
  • Category 4,    2ms:  2.8 MBit/s
  • Category 5,  10ms:  2.0 MBit/s (no, not a mistake, it's slower)
  • Category 6,    2ms:  5.7 MBit/s
  • Category 7,    2ms: 11.5 MBit/s

For the nitty gritty details see table 5.1g. To get to the raw speeds quoted above, divide the maximum number of bits per transport block by the transport block length (the tranmit time interval). Example: Category 2 mobiles can send up to 14484 bits in 10ms (TTI). That's 14484 bit / 0.001s = 1.448 MBit/s. Note that there are 2 TTI lengths, 10ms and 2ms. For higher speeds, shorter TTIs are needed and the 2ms block has to be used for maximum speed.

The speed increases are achieved by lowering the Spreading Factor from 4 to 2 (i.e. how many bits (or chips to be exact) are required to encode one user data bit on the air interface) and by increasing the number of simultaneous code streams (from one in category 1 to four in category 7).

A strong word of caution: Like for the downlink, these values are top speeds. The higher the maximum speed the less likely it is a user will experience them in a live newtork as the signal conditions under which they are achievable are only available in an ever shrinking part of a cell. Most users will not see such high speeds, especially when they are indoors and have no line of sight to the base station antennas.

HSPA+ Speeds in Which Areas

I keep mentioning in posts on HSPA data speeds that one has to be very careful when using such numbers as they represent the theoretical maximum that is only reached very close to the base station. In various presentations, one can often see graphs where speeds are shown over the percentage of users, i.e. what speeds 95% of the users are experiencing to the speed only experienced by 5% of the users. But what does that mean when plotted over a geographical area?

Ericsson has done a great job of visualizing this in their 1/2009 edition of the Ericsson review on page 8 figure 2. I can't reproduce it here, but I encourage you to follow the link and check it out for yourself, it's very insightful! As you can see in the 3 pictures, which represent various stages of the HSPA+ evolution, the direct benefit of adding higher top speeds gets more and more limited the higher the speed.

What the pictures don't show, however, is that subscribers not enjoying the best radio coverage do also benefit from higher speeds, in a different way though. By having at least some users in the zones where data can be transmitted faster, users under ordinary conditions benefit because the overall cell capacity has increased, as some uses can transfer their data faster, thus leaving more time to serve other users.

Great visualizations, thanks to Stefan Ström, Dirk Gerstenberger, Johan Bergman and Fredrick Gunnarsson for the article!

Shrinking Base Stations

The picture on the left is an interesting example of how the latest generations of GSM and UMTS base stations are replacing legacy equipment in the field. Previously there were two base stations at the same place (one for GSM and one for UMTS) taking up most of the space of the gray plateau. The small base station now in place is about 1/5 to 1/6 the size of the two original cabinets and includes GSM and UMTS elements plus the equipment necessary for the microwave backhaul.

The refresh cycle is also interesting. I would estimate that the original GSM base station was in place for around 6 years and the UMTS base station for around 3 years. With the original GSM base station probably having been end of live, I guess it was a smart move to replace both at the same time. What I also noticed was that for two of the three radio sectors, the original two antennas, one for GSM and one for UMTS, were replaced by a common antenna including remote electrical tilt (RETA) for later fine tuning.

And finally, compare the size of this GSM and UMTS base station to much bigger VDSL cabinets, which are deployed with the same or even higher density as wireless base stations these days (I estimate their distance to be about 500m). Interesting to contemplate about the synergies and deployment costs.

Playing with the Digital Dividend

The German government has recently announced its intentions to reassign analog TV spectrum between 790 and 862 MHz, the so called digital dividend, for providing broadband wireless Internet to rural areas. With this intention it follows other EU member states and it looks like we’ll get another harmonized band in Europe pretty soon.

I was hoping this band is similar to the digital dividend band in the U.S., which was awarded to a number of operators some time ago. Unfortunately EU digital dividend band is about 100 MHz above the US 700 MHz band. Too bad, it would have been nice to have a common band, so a single device could have been manufactured and used in both parts of the world. It looks like we won’t see economies of scale and global roaming for this band either. And this would have been direly needed as the digital dividend band at least in Europe addresses mainly rural communities due to its propagation characteristics, a minority application in the global wireless game anyway.

The article announcing the re-assignment of the digital dividend band in Germany also contains a quote from an industry representative saying that if the German government goes ahead quickly, deployment of Internet access in this band could start in 2010. Hm, I wonder what kind of network they would use!? It’s unlikely it’s going to be either HSPA or LTE. For both, this frequency band is not yet standardized (see 3GPP TS 25.101 and 3GPP TS 36.101). A long way to go, I seriously doubt 2010 for those two systems.