Category: Uncategorized

What do I do on a short stopover in Japan? Right, I check out the 3G networks there. After all, NTT DoCoMo was the first network operator to launch UMTS (FOMA). Also, they have no GSM network to fall back to so they need to think a lot harder about 3G indoor coverage and ubiquitous coverage on trains, etc., so voice calls don't drop. In addition, if the 3G network works well in Tokyo with its high population density there's no excuse anywhere else. And indeed, the network worked very well during my stay and the setup was unlike anywhere else.

On the way from the airport small screen web surfing on my mobile worked just fine and I never lost the network. Also, I never lost the network indoors. In a shopping mall I always had a strong signal and only one cell, so it's likely there was dedicated indoor coverage and no repeater bringing in the signal from outside.

At first, it looked like DoCoMo had only one carrier on air in the Ginza district, as no inter-frequency measurements were configured by the network. But when transferring data, the UE was immediately put on another carrier frequency while it was in Cell-DCH state. I observed two additional carriers. Here, the UE was kept for around 10 seconds after the last piece of data was transferred, a value seen in many other networks as well.

When no more data is transferred, the mobile is returned to the initial carrier and kept in Cell-FACH state for around 90 seconds. Not very energy efficient and much longer values compared to those I've observed in many other networks which range from 10 to 30 seconds. (Hello Fast Dormancy!)

If still no data was transferred during that time, the network sets the connection in Cell-PCH state for around 2 minutes after which the connection is put into the idle state. Again quite a surprise for two reasons. Why a 90 seconds FACH phase when the networks puts the connection in Cell-PCH afterwards? And why only 2 minutes in this state afterwards? I am sure they have their reasons but I can't quite figure them out. Most (of the few) other networks that use Cell-PCH I have observed in Europe so far quickly put the mobile to the Cell-PCH state, for example after 5-10 seconds in Cell-DCH state and a further 5-10 seconds in Cell-FACH state. Afterwards, mobiles are kept there for long durations, 30 minutes not being uncommon. Some networks even use the URA-PCH state if a mobile moves between cells during that time to reduce the amount of required signaling. Quite a different strategy.

Ah yes, and then there are voice calls. I made a couple of calls and they were all handled on the basic carrier.

Softbank also has a 3G network in Japan and it would have been interesting to see what they are doing. Unfortunately, no roaming agreement was in place so my mobile would not roam on that network. Well, next time then…

Here's a press report I picked up recently from Kineto, who've ported their UMA protocol stack and application to Android Froyo. This may sound a bit cryptic but it has interesting implications for Voice over LTE.

For those of you who can't place the term UMA (Unlicensed Mobile Access), it's basically GSM voic calls over Wi-Fi. On the mobile device, all layers of the protocol stack of GSM are reused with the exception of the lowest layers where the UMA software introduces a switch so phone calls and signaling are either using the GSM infrastructure or a Wi-Fi access point that connects to the Internet. For the user its completely seamless. On the network side, the base station and the base station controler is replaced by a UMA gateway controler that connects to the Internet on the one side and to the network operator's core network on the other. To the core network the gateway looks like a GSM base station controler which makes UMA also transparent to the core network, i.e. no software or hardware chagnes needed there at all.

So UMA on Android, what does that have to do with voice over LTE? Well, as you might be aware of, there's no built in voice telephony functionality in LTE. That's why a number of different options have been suggested, one of them being Voice over LTE via GAN, or VoLGA for short. Now GAN (Generic Access Network) is just a synonym for UMA and that already tells most of the story. Volga replaces the Wi-Fi layer of UMA with LTE.

In other words, to close the circle, there's only little work required to modify the UMA Client for Android to use not only a Wi-Fi network but also an LTE network once Android LTE devices become available. With a client available for Android, operator based Voice over IP has arrived in the smarphone space (beyond RIM) and on an operating system that will play a major role in future LTE devices. And the work to make the software work with LTE is almost trivial since both Wi-Fi and LTE are IP based networks. So 99.9% of the software don't care if Wi-Fi or LTE is used for the transport of the IP packets that contain the speech data of an ongoing voice call.

One additional thing that requires a bit more work is the handover of an ongoing voice call from LTE to GSM (or UMTS) when the user is at the coverage limit of the LTE network. But like UMA, which is a 3GPP standardized technology, 3GPP also comes to help out here as there are already methods defined to do just that. The functionality is referred to a Single Radio Voice Call Continuity (SR-VCC). Specified in a generic way it can be used for IMS and also for Volga and potentially other solutions in the future.

While it takes a bit of work to implement SR-VCC into Volga, its fully specified and well worth to be done, as LTE – GSM handover of voice calls is a network operator's biggest asset to stay relevant in the voice call domain in the future as over the top voice applications have no way of doing that.

Here I am again in a hotel in the US, the Hilton in San Jose to be exact, and I am glad that I have quad-band 3G stick and the AT&T prepaid SIM with me, as the Wi-Fi/fixed Ethernet connectivity in the hotel just doesn't cut it. Available for 10 dollars extra a day, I just get a meager 150 kbit/s through the line with highly variable delays and packet loss. Even at 3 a.m. (hello jet lag…) the speed is less than 1 MBit/s. No matter how many MBit/s your DSL line is capable of, dear Hilton manager, it's far from enough. We are in Silicon valley, so heavy Internet use should not be surprising!? So I vote with my feet. No money for you for the Internet access, it's gone to AT&T. Not that their 3G network is anywhere close to the best I've seen with the 2 MBit/s I get, but way beyond what comes out of that Ethernet jack in my room. Reminds me a bit of the stick meeting post I wrote some time ago. Let's see how many 3G sticks are used in the meeting today in the US.

Just recently I was, and actually I quite am, quite extatic about having reached 11 MBit/s throughput in a live 3G network over here in Germany. But the specs are already two orders of magnitude ahead.

In a recent 3GPP RAN meeting, a new set of LTE-Advanced UE categories have been specified. The highest one, Category 8, specifies a transfer mode for 1200 MBit/s downlink and 600 MBit/s uplink transmissions. That is, of course, if we'll ever see devices with 8 receiver antennas and networks that have base stations with 8 transmit antennas per sector. Also, a channel bandwdith of 40 MHz is required. Each parameter on its own is already quite a stretch when compared to the current state of the art. But then, only ten years ago, we were creeping along with wireless speeds of just a couple of kilobits per second. Two orders of magnitude less, actually.

For those who want to have a closer here's a link to the 3GPP RAN WG1#62 report. In the report, search for R1-104944.

While I'm personally a bit sad that non of the new Symbian^3 based phones Nokia presented at Nokia World recently is a vertical slider with a traditional T9 keypad there's one thing that makes my international roaming heart jump: All devices presented, namely the C6, C7, E7 and N8 all come with a pentaband 3G receiver included. In other words, while my current single band UMTS device has to fall back to GSM when roaming for example in the US or Canada, these devices should work pretty much everywhere you go. Also good for Nokia as that significantly reduces the number of device variants. For details here are the product pages for the N8 and E7.

In some countries in Europe such as Germany, for example, there are quite a number of offers for daily 3G Internet access now in the range between €2.50 to €5.00. Especially €2.50 offers sound quite cheap, almost like network operators giving their services away for free. But how much is that compared to the daily fee of a DSL line at home? Let's compare with a DSL line that costs €30 per month. That would be €1.00 a day. In other words, there's still a premium for wireless in the daily charge. So while it sounds cheap, summing up €2.50 over 30 days brings you to €75. That's quite a sum so it quickly pays to go for a monthly option if mobile Internet access is required for more than just a few days per month.

Every now and then I see in computer magazines or tech websites some network speed measurements (including my blog). Often, however, they are based on single measurement experience or maybe based on using a notebook in a few areas in the network. While this might be good for telling the performance in those places it doesn't really capture the overall capabilities of the networks. For this, so called drive tests are required and for comparisons between different networks, measurements must be performed in the different networks simultaneously. That's quite a challenge but the resulting data is well worth it. Here's a link to a recent article in a the German computer magazine "Chip" that has made an extensive test of the German networks. Google can help to translate the pages for those non-German speakers of you. Highly recommended for the background information and the interesting pictures of the measurement equipment and the results. If you know of similar tests having been published in other countries, please let me know!

It's good to see that the number of prepaid SIM offers in many countries that can be used for 3G Internet access keeps going up and the Prepaid Wireless Internet Wiki keeps growing and information is constantly updated. Also, I keep noticing that prices are getting more competitive. For my two weeks of vacation in Italy this summer, for example, I used a SIM card from mobile operator WIND which I already used during previous trips and noticed that prices have further gone done. For my two week stay, the 1GB / month data option was quite sufficient which now only costs 9 Euros + a 4 Euro activation charge. For longer stays the 10 GB / month option for 20 euros might also be an alternative. Compare that to what some hotels charge for Wi-Fi per day…

Hm, have a look at the design of the 3-year old Nokia N82 on the picture on the left. Those rounded edges and metallic look are quite similar to the general exterior design of a recent popular product!? Except, of course, that the frame is not used as an antenna. Suddenly, the phone is twice as beautiful as before…

Ever since I first visited the Fraunhofer FOKUS institute almost exactly two years ago, I've remained in contact with the researchers there as they are working on interesting research projects in the wireless core network and IMS domain. Next month, on the 14th and 15th of October, they are hosting a follow up to their IMS conferences in the previous years called "Future Seamless Communication" or FUSECO for short. They've got a long list of interesting topics and speakers for their conference both from the industry and academia, something I always enjoy as both sides can learn a lot from each other. If you are interested in wireless core network topics it's definitely a conference not to be missed. Here's the link for a closer look. Enjoy!