Tuesday, June 27, 2006

A long talk with Mikael Bäck, Ericsson's HSDPA honcho

It has been a while since I spoke to Mikael Bäck, Vice President of WCDMA Radio Networks at Ericsson. An interview article was published in my newspaper (in Latvian), but a series of unintended circumstances prevented it from being published by one of my major freelance outlets. So as not to let this go to waste, I am presenting, here, a rough cut transcript of our talk in Stockholm some weeks ago.

What is HSDPA?

3G in the beginning introduced a wider radio carrier of 5 Mhz instead of 200 Khz, which gives us totally different possibilities. The first step in wideband CDMA doesn’t give us the full data capabilities, it is limited to 384 Kbps video and voice. HSDPA is the most efficient radio high data carrier that we have and we can bring theoretically even higher data rates that 14.4 Mbps over this frequency. That is up in operation today, better than anything anyone has been able to use in the world in the street. So if you compare it, the only other possibility to get these data rates is through wireless LAN in the office, which is limited.

What does this mean for business users? SMEs have internet in the office?

We believe it will be used in many different ways. The first step, partly because it is a good business case and is the easiest case is that people will be using the PC card to access internet from the laptop only. They will do what they do in the office without the cable.
Then what will come soon, is you will be able to mix voice and data in terminals, you will have a fixed terminal in some countries where in small offices you replace fixed for data. All this in combination means we will be able to address an extreme variety of needs without the operator having to invest in specific systems, as would be the case with specific systems for mobile TV and WIMAX. I think we can do a lot of things with only one infrastructure investment for the operators, and that is very much behind this.

How does it work?

It is a software upgrade from Wideband CDMA. For new operators, almost everybody is buying HSDPA ready equipment.

So, setting up a brand new operator and infrastructure, I would get HSDPA in the package?

That’s the way it works for all the new ones, if that would be in Eastern Europe or South Africa. It comes with HSDPA as the basic feature now.

How much uptake is there of HSDPA in the real world?

If you look at the uptake, I don’t know what percentage of the installed base has been upgraded, but it is a pretty good part now. It takes some time, however. We forget the marketing campaigns the operators must have before we get real users. In Europe, maybe Mobilcom in Austria is the biggest. Elsewhere it is small. Here is Sweden we don’t have it.

Why is TeliaSonera not doing it here but at their subsidiary companies Estonian Mobile Telephone and Omnitel in Lithuania?

That’s difficult to judge. In Sweden, it is a large area. In Sweden we will start in the big cities and EDGE is a feasible alternative for a lot of rural areas that are much larger in Sweden than in the Baltics. And it may also be that the competition between operators in Sweden is not so hard.

Once the phones arrive, what will be the content and the applications that are tailored to HSDPA?

Something that is a big thing already in 3G and that will take off is mobile TV. Different kind of tailored material. Either normal TV channels or content that is tailored for mobile, streamed or download pay per view or view or by different kinds of subscriptions. That is one of the things you already see in Western Europe. Quite a lot of operators have relations to a parent company that is involved in a media company. That will start quite early. 3 in Italy has no fixed line offering so they target pure competition with the DSL networks, they offer 1.8 Mbps for EUR 9.00 and will upgrade it to 3.6 Mbps and they call it ADSM. For them that’s pure competition with the higher priced fixed line. So you will see different types of alternatives coming out.
One of our customers in Japan, Vodafone KK, was recently bought out by Softbank, the biggest player in DSL in Japan and also the owner of Yahoo! Japan, so they have very advanced plans to introduce a lot of the services you see today on the internet, including voice over IP. Again we might see a trend that some of the first applications will be around in countries like Korea, Japan before they enter Europe which has traditionally been a bit slower on new services

Now that you’ve compared HSDPA to DSL, aren’t you facing the problem of being a supplier to both the fixed line operators and to the potential “cannibals” of fixed line DSL?

That’s part of our role. If you want to be one of the biggest suppliers in telecoms, it is totally unavoidable that you supply to competitors. In reality, you will have many different types of systems, WIMAX and such. It is very difficult for us to say that we will stay out of this part of the competition. We will have the easy cases, the 3s and the Vodafones that don’t have a fixed line offering. For them the positioning might be easier. But a big part of our customers, that’s the KPNs., the Telefonicas, the Telias, are very strong on the fixed and very strong on the mobile side. For them it is a very interesting internal debate on how are they going to position this. And very often, they start to position this not as cannibalizing DSL but as the businessman’s portable solution on the laptop that would not compete as much with the DSL lines, while 3 is competing directly with DSL.

You say HSDPA will be used for business. How much are you working with enterprise software companies and how will you deal with security if you have people working in the park with the SAP or Oracle ERP system and sensitive data?

We have a number of different things that will happen. If you take the device as such, you have the industry, driven by the Intels and possibly the AMDs, where today wireless LAN is integrated into the laptops. Before the Centrino, where Intel integrated wireless LAN into the motherboard, when you had separate PC cards, no-one believed that you would have wireless LAN in more than 40 % of laptops. Then it came as part of the Centrino platfom and with Intel holding an extremely large part of the laptop market, suddenly the uptake of wireless LAN on new laptops was 94 % instead of 40 %.
From my point of view, what we have tried to do with Ericsson mobile platforms in the same way as Qualcomm and other companies do, is to try to get HSDPA as integrated with laptops are wireless LAN has been. And that will come in various steps, starting with the PC card, which is the most simple. But there will be a market, and we will help to create it, where HSDPA is installed in the laptop as is. You can already buy such laptops, but the price is fairly high.

Who is making these laptops? Do you have an alliance with anyone?

If you look at what has been publicly announced, there have been announcements from Sony, from Dell, there are operators like Cingular, Vodafone, T-Mobile that are bundling, more or less hardware with the laptops. There is a small company called Flybook selling fairly expensive stylish Italian designed and today you can order one of those laptops with integrated HSDPA, not a PC card. It would be very strange if we couldn’t get it into every brand. They only problem is that if we can’t get the cost down, it will only be in the higher priced models.
Then, when it comes to how it works with the laptop, we are working with Ericsson Enterprise that is doing applications and clients for the P900 and those kinds of phones, but we will be working with that environment to create a secure environment, a simple password and SIM card handling environment. This will be important, since today the solution is a little bit too difficult. We have a separate key generator and you have to dial in a special number and get another code.

So today HSDPA equipped laptops have enterprise level security?

Yes, we use it here in Ericsson, I can use it. But it is still a bit of a hassle to set it up, because you have to create the secure tunnel. It has to be a bit simpler to be a hit in all corporations. But it is approved in all our own internal processes.

How can others do this?

A secure network can be run over HSDPA. The IT department would typically set it up, but you can purchase the solution from Ericsson or the other typical players you have, the laptop vendors and the Microsofts. That is part of the process why in some places this is very slow, these are very static parts of the organization.
The solution and the problem is the same as with wireless LAN. We don’t allow open wireless LAN at Ericsson, because then someone standing outside the building could use the network. I set up a secure tunnel and the solution is the same whether I am using wireless LAN or HSDPA.
I think it is a very impressive service, it runs at 1.8 Mbps now and will soon run at 7 or 8 Mbps. It is not as fast as the wireline network here at Ericsson, but certainly as fast as some of the fixed network DSL you could access abroad and it is enough for the kind of work I do and attachments I might need to send or receive.

What is the situation on the handset side? This is what is expected to generate all kinds of content viewing and revenue…

The first handsets will be available very after the summer, our guess is in September, that is the information we have now. We have them in the labs only. You can’t buy them on the street.

When Estonian Mobile Telephone announced its HSDPA service, it mentioned download speeds of 14.4 Mbps, which some observers received rather skeptically. What is the evolution track for HSDPA speeds and what will come after HSDPA?

If you take what we have defined as HSPA today – the D has disappeared because that refers to downlink and we have an evolution for both the downlink and the uplink– and if we could build as fast as possible according to the standard, it allows 14.4 Mbps on the downlink and up to 5 Mbps on the uplink. So it is up to us vendors to design the equipment. Today our system can run 3.6 Mbps with the release that is out there, in the Baltic States, for example. The PC cards and the terminals can run 1.8 Mbps. Then in the next release that will come out around Christmas will have the possibility to run in the network at 14.4 Mbps. It seems like the biggest part of the devices will go for 7.2 Mbps and maybe not higher for a while, maybe with an enhanced upload.

When you talk about devices, you mean PC cards, phones. But are there any other devices coming out, perhaps TV-like tablets?

Not so much has been announced yet by suppliers, but in reality if we have a market of small modules performing HSDPA operations, there is nothing preventing this device from being the communication device between tablets, between iPods or whatever kind of device that communicates and that is up to the device makers. And it depends on the price point coming down.

Using iPod as a general term for all these music devices, can we see iPods with HSDPA coming out soon?

Absolutely. If you look at the pure silicon part that does HSDPA, it is small enough to be inside anything. Then it is a question of how it will be sold, in small plastic modules or whatever, Maybe it could be difficult to integrate in the small iPod Nano size, but there is no reason it shouldn’t be in any kind of device.

So you could have an Mp3 player that is always on and searching for any kind of entertainment that is available?

That is my guess for one of the killer apps if you don’t take the normal broadband laptop broadband market. Being always connected to music, be it iTunes or the like, is an obviously very interesting market.

What about TV on demand, or TV that comes on when it detects something of interest? Say, during an election campaign, a candidate of interest is appearing in a news spot?

There is TV up and running and we have been doing tests to see which kind of data rates we can get and how that will affect TV service. At our recent event in Rome we showed a quite big, normal 32 inch display where we streamed two videos one over 384 kbps over wideband CDMA and the other HSDPA and it was an extremely good picture on the HSDPA.

So, assuming coverage, you could go out on a sailboat and watch HSDPA TV through a phone and a 20 inch LCD screen?

You will do that! But when you are talking about only downlink, you will always have competition from normal TV service. What will happen is that interaction will come more and more into these type of services. A device that cannot communicate will be less and less interesting. That will be one of the other killer apps.
One of the most important battles in the industry is now going on in Australia and can be followed on the internet because it is extremely public. What Telstra is doing is using HSDPA to cover the desert in the middle of Australia with extremely few sites. We provide them with a system that can run, as we have demonstrated in the desert up to 80 kilometers and can run up to 200 kilometers from one site, basically normally over water. And we can take care of the delay that would come over that distance.
In the desert from a high mast we can reach 1.8 Mbps over 79 kilometers distance.

So basically you are competing with WIMAX?

Absolutely, from that point of view. We are doing that in a good way, because WIMAX doesn’t exist, which is always a drawback. The WIMAX we have today are the pre-versions and not what will be the real WIMAX coming out from Intel in, say 2008. So we have a very interesting market window.

Is there a name for this service? I understand WCDMA runs on frequencies that don’t propagate very far?

What Telstra is doing is running this on 850 Mhz, which is better. But in many areas, as long as you have not really line of sight, but not many things in between, or you use a high mast or a high mountain, then you can propagate quite far with the 2100 Mhz. maybe not 200 kilometers, but if you reach 80 kilometers, it is an extreme distance.

In Latvia and some other East European countries, we have CDMA 450 operators who say they can deliver wireless broadband at 1 Mbps or under optimal conditions, over 2 Mbps. Where does this system, EV DO, fit in with HSDPA?

It (HSDPA) is a wider band and the laws of physics decide how many bits per second you can get out of a piece of spectrum. The CDMA is only 1.25 Mhz and the wideband CDMA is 5 Mhz. So the CDMA systems will be limited unless you add frequencies. The highest possible data rate you can get with CDMA is around 2 Mbps, while it is theoretically around 40 Mbps for wideband. If the theoretical limit is 2 Mbps, then you will have difficultly getting 1 Mbps unless you are the only user on the system.
My view is that realistically the 450 band is very difficult, it has good propagation. It was used in NMT for voice, but for any kind of data services, it is very difficult to use.

In Latvia we have Triatel, the CDMA operator, selling these solutions out of a box, voice and internet, and allying recently with Lattelecom to build out the network with wireless in the countryside. Ar you saying that HSDPA is the better solution for providing really fast broadband in the countryside?

The problem with CDMA 450 is that it will work perfectly well for voice, it will work as well as the (analog) NMT 450 system did, but with smaller telephones, It will be very good for voice, not for other things. Unless you create something very new, the 450 band is not good for data. In Europe you use the 2100 Mhz, the 1800 Mhz bands, the 900 and 850 Mhz.

So if the telcos, such as Lattelecom in Latvia, want to offer wireless broadband in the 10 Mbps range, as they are offering in the cities, they will have to turn to HSPA?

That is my view, If I look the things I have been working with with wideband and before wideband, HSDPA is the absolutely the single item that is creating interest on the market among both the end users, the operators, the content industry. I think this will be looked back upon as a very important step for going away from cable broadband services.

So you are saying that when it is all developed, networks, terminals etc., it will be as fast broadband as anyone with normal consumer or enterprise needs will require?

That is my view. In reality all types of DSL services are limited by the distance to the station, and in most places you will be limited to around 10 Mbps if you are not living close to the station. So the only way to get much higher rates is fiber to the user, which is very expensive. So we will see different behaviors on different markets. In the Nordic countries, it is difficult to see this kind of fiber development, in the US you might see that trend.

What is over the horizon after HSPA? How fast can radio transmit data?

We can go very high and what we are doing now is introducing the next step in standardization that we call long-term evolution where we will use wider bands, which is realistic in Europe, up to 20 Mhz, which means you will be able to transmit 100 Mbps, in reality up to 200 Mbps. You can go even higher with wider bands, but those are hard to find, frequencies are a limited resource.

So if there is going to be an HSPA 2.0 it won’t go higher than 200 Mbps?

I think that will be a practical limitation for quite some time. And it is difficult to forsee needs that you will not be able to serve with that kind of capacity.
A problem we are working on today is latency, something that happens on GSM and GPRS can be hundreds of milliseconds and affects such applications as gaming or protocols where you have a lot of acknowledgement will have performance lags. What we are doing with HSDPA is getting those latencies down to 75 milliseconds and we may get it down to 10 milliseconds, which will open up a whole world when it comes to voice over IP, push to talk, where delay is an issue.

You can run VOIP on HSDPA today?

Yes. It works well.

So we are back to creating more cannibals, aren’t we?

It is up to price model and the operator. If you charge a high amount for the voice and low for the date, you create that kind of cannibalism issue.

Ok, but if you want to sell lots of data services at attractive rates and you have that other company across the Baltic Sea (Nokia) releasing phones with WLAN capability, what is going to happen to GSM voice revenues?

Of course you see already today a trend toward lower cost for voice and different types of bundlings. You have that kind of development whatever you do. In Sweden we have had drastic price changes in just half a year.

But if the network sees many phones as just IP devices, then your call to another one of them will simply be part of some flat rate data service…

This is a fairly complicated market. If you look at the internet and the enterprise networks, then voice is a very small component in a data network. On the mobile networks, voice is 99 % of the traffic. Operators will try to block Skype type of services for a while, then they will come to their senses. But it will not be a technology issue, it will be how operators package these things, how they offer voice over IP.

Still, thinking about HSDPA, I am reminded of the 1980s Gremlins movies, where these nice fuzzy creatures got let into an office building, then they got wet and turned into these raging little monsters that tore the place apart. Hasn’t HSDPA put the Gremlins into the GSM building?

It would be very strange if certain things that happened on the fixed networks didn’t happen on the mobile side. It is so that most operators don’t want to lose control and become a bit pipe. They are opening up the bit pipe but trusting that there are no devices that make it easy to use, like it is impractical to use Skype on a laptop, you don’t walk around with it and talk. The first devices are expensive, so there will be some time to fiddle around with the strategies and these first devices won’t hit as hard as they did on the fixed lines. But this is big question that operators are working on.
Also, the application development for mobile phones is not as fast as the extremely fast application development happening on the internet side. You will have two markets meeting. If you go to an all IP world on the terminals, which has to happen, of course there will a lot of players like the Googles and the Yahoos that will want to get into this business the same way mobile players are trying to replace the fixed line DSLs. We have been talking about convergence, but it has not happened, it has mainly been bundling of different offerings. Here it is very obvious that this is convergence. You will have the same type of services and then you will have some really strong tools in the mobile environment, that you will have mobility which you are not used to from the fixed line side. We are working a lot with IMS to create an environment where you have presence information, so you tailor your offerings a lot more, location based and so on.

So IMS is important for HSDPA?

IMS is formally a separate thing but it comes together very much, IMS if you don’t have a good bearer will not be as strong. Of course IMS is for both the fixed line networks and the wireless LANS and HSDPA. But for me the strong correlation is with HSDPA.

So if you are at home watching a football game on TV and you are called to the office and have an IMS compliant device, you can continue watching it on your mobile in a cab…

That is one of the extremely strong offerings that could come from us to operators and to the end user. But it is also fair to say that this type of development will take quite a long time, maybe not for us, but this is quite a big change if you look at world most operators are living in, how they sell phones and distribute content. It is quite a big change and it will take some time. But one thing we have seen with mobile broadband and with mobile telephony is that if you want to have a fast uptake, it is a very good step to take a service that is existing already, such as telephony was existing in the fixed line network and move that to mobile telephony, because you have an immediate understanding of what the service is for me as an end user. The same goes for mobile broadband, if you take what you have in the office or on the home DSL and move that to a mobile environment. People understand how to use it and why they want to have it. The same could go for TV as well. Whatever the TV you have at home, or the IP TV coming on now, you would be able to take it with you when you walk out of the home. You understand why I want that and so on..

So that is it, not the usual blog format.


Anonymous said...

Great stuff! I will give it some time and come back with comments later.. I definitely have a few, not in the last place about 450 MHz not being suitable for data. It's Ericsson Corporate propaganda :-))

I will create a username as well . It´s about time since I have been posting / commenting several times here and I aim to continue with that.

So far...
More to come.

Juris Kaža said...

Those assertions about cdma450 left me a little skeptical. However, doesn't Ericsson make a lot of cdma stuff for the US? EV DO is supposed to be quite fast in its newer generation versions, so said the Qualcomm guys who were in Riga some months ago.

Bleveland said...

Well Ericsson acquired major parts of Qualcomms infrastructure division 1999. http://www.findarticles.com/p/articles/mi_m0UNZ/is_1999_March_29/ai_54242672 That's how Ericsson moved towards CDMA. The CDMA stuff for the US is "old" CDMA according to the IS-95 standard as well as CDMA2000 for 800 and 1900 MHz. In general CDMA2000 is just upgraded CDMA for several frequency bands, mainly 850 and 1900 MHz. CDMA450 is CDMA2000 technique at 450 MHz.

It seems that the people at Ericsson have been having difficulties to make up their mind about CDMA450. In 2004 Ericsson's CEO Carl-Henrik Svanberg "launched" CDMA450 during his first trip to Russia as CEO for Ericsson. It was a part of Ericsson's Expander product portfolio; mobile systems for rural areas and growing markets (in those days with one billion mobile users world wide called the "Next billion user market" - by evil minds sometimes called "mobile telephony for the poor").
Since I happen to know that you understand Swedish very well Juris, you can read about it here: http://www.nyteknik.se/art/33065 (sorry for those who do not understand Swedish). A year later, early 2005, Ericsson went a step further and outsourced development of modems, PC-cards etc for CDMA450: http://www.nyteknik.se/art/39176
However, as far as I am aware of, they seem to have missed the train and not many commercial CDMA450 networks seem to be equipped with Ericsson stuff.
I am not sure if they still offer CDMA450, but I did not find much about it, so I guess not. They even (re)continued the old GSM450 development again. Once started in 2000 it was put on hold due to lack of demands from the market. Yeah, by that time Ericsson still made stuff that nobody asked for :-))
Anyway, GSM450 is back on track since fall 2005 which makes it even more likely that CDMA450 is dead as far as Ericsson is concerned. The interview with Mikael Bäck and his fairly negative comments about CDMA450 seem to confirm that. Probably just a business case issue. Rather low volumes and low-cost demands plus not being the first offering it made Ericsson not the supplier of choice. That could explain the kind of reaction "It's inferior technology. Ericsson does not believe in it". Just a guess... Here is the link about GSM450 (again på svenska): http://www.nyteknik.se/art/42700

The following text is copied from an article about Ericsson Expander written by Ericsson (http://www.ericsson.com/ericsson/corpinfo/publications/review/2004_02/files/2004123.pdf). Remember what Bäck said and then read this (from 2004):

....Thanks to better propagation, products operating in the 450MHz frequency band can cover rural areas at a fraction of the cost of networks operating at higher frequencies.
For example, a base station operating in the 450MHz frequency band effectively covers
• 100% more area than the same base station operating at 800MHz; and
• 500% more area than the same base station operating at 2.1GHz.
Rural operators can thus significantly reduce their capital and operating expenditures.
The solution is of particular interest to NMT450 operators who want to continue using the 450MHz spectrum, and to operators in countries where CDMA450 technology
is being considered for either rural or fixed wireless access (FWA) coverage....

Not exactly 100% what Bäck said... So much for 450 not being suitable for data just because it is in the 450 MHz frequency range. B*llshit in other words. Bäck is however right in many (other) ways.

It is true that HSDPA has much more potential when it comes to speed and capacity. Not in the last place because it runs at frequency bands with a lot of frequency space (= many channels). The capacity on 450 MHz is indeed very limited. As I mentioned before in an earlier comment CDMA works fine for data as long as there are not to many users connected to a base station. With many user data rates decrease dramatically. There are just a few channels available in the 450 band, so increasing capacity for instance in Triatels case is hardly possible since one channel will be used for EV-DO and one or more are absolutely needed for voice. It seems that Triatel has 4 frequencies / channels (Telekom Baltijas 1.9 MHz band width = 1 channel and Radiokoms 4.5 MHz = 3 channels). So that leaves not much. Then again in rural areas there is probably no need for so much more and CDMA450 might work just fine there. Applying "fair user policies" and offering "best effort" service as well as blocking for instance p2p to avoid continuous heavy data load is probably what Triatel is forced to do in order to avoid major network black-outs. Not everybody will read email and surf the web simultaneously. Loading a web site will generate a short peak in the network load, but statistically since these peaks -most likely- will not occur at the same time it will make the network look fast for each individual user. Even though there are quite a lot of users on-line.

As for data rates 1xEV-DO has a maximum theoretical data rate of 3.1 Mbps (download) and 153,6 Kbps (upload). The upload limitation can be a real deal breaker. The newer version the Qualcomm guys were talking about is probably 1x EV-DV which addresses the upload issue and gives 3.5 down and 1.8 Mbps up. The next phase will be aggregation of several channels and the introduction of OFDMA within CDMA as well as more advanced modulation schemes. This will allow download peak rates of roughly 70 Mbps and uploads of 25. The next step will be the introduction of dynamic channel bandwidth wich is believed to give peak rates in the ranch of 200 Mbps.
Ehh... wait a minute.. wasn't that exactly what Bäck said about HS(D)PA? So by that time WCMDA (UMTS) and CDMA2000 might have merged together. Ericsson is already building their CDMA2000 system on the same platform as WCDMA (HSDPA). Since also WiMAX is using very similar modulations schemes in the end they all may very well end up with technically the same solution :-))

BTW at least one network infrastructure supplier for Triatel is Canadian Nortel (there might be others (?)).

OK. Enough (more than) about it.

Anonymous said...

Here’s a Broadband Video that will show you how to check availability by postcode, how perform a broadband speed test and where to find broadband forums to answer your questions. There are also offers for TalkTalk Broadband.

Here is a broadband beginners guide and below are common broad band questions;
What is Broadband?
Types of Connections?
Connection speeds and download limits
How to compare ISP’s