February 2008 Archives

The following interview is with Dean Bubley who is the the Founder of Disruptive Analysis, an independent technology industry analyst and consulting firm. It took place on the 29th January 2008 - yes I know I am a little late getting it out the door! Dean will be delivering one of the keynotes at eComm next month.

The run time is 1 hour and 2 minutes.

Lee: So today I'm sitting -- I keep saying that actually, I'm sitting with blah-blah-blah, and I'm not sitting, because you're there and I'm here and we're on Skype. So today I'm chatting with Dean of Disruptive Wireless. Dean, can you tell me why you think eComm is important?

Dean: I think it's going to be a really interesting event, because every day I hear a huge amount about new and potential ways of using communication. And to be honest, I would rather filter them myself. As an analyst, my job is to try to wade through reams of possibilities, of pitches from all sorts of different companies, and being able to hear it outsourced and decide what I think is real, what I think is reasonable and what I think is unrealistic is very valuable. And I think it should be a fascinating events to actually see things put into context against each other. And there may well be a bunch of things which surprise me, positively and negatively maybe.

Lee: Okay. So, Dean, you're coming along as some kind of filtering critic, then. So I look forward to that. You are going to be speaking on a topic entitled "Who Controls Wireless Access: Carriers, Internet Players or the End User?" Can you give me a very brief outline of that?

Dean: Yeah. I mean, essentially this is looking at the realistic change for some of the more utopian ideals of mobile and wireless communications to become real.

Why utopian? I mean, there are an awful lot of people who refer to openness, who are looking to regain control of their own sort of mobility rather than leaving it in the hands of licensed mobile operators and carriers.

And up to a point, I can certainly understand why that's desirable; it's never good to have an unnecessary middleman between you and what you want to achieve.

On the other hand, we're in a practical hand. There are areas where services make sense. There are areas where things like licensed spectrum make sense, as well as unlicensed spectrum.

The fact is that the world has benefitted from several billion people using cellular telephony, for example, despite the fact that it's being controlled in a fairly rigorous way by the mobile operators. An interesting question is whether that will continue as you go forward with the advent of new wireless technologies, whether it's Wi-Fi meshes or WiMAX and various 4G and 3.5G technologies coming fairly soon. And then on further on the horizon, you've got things like software-defined radio. And there's a whole set of issues around policy, technology, end-user behavior and regulation, all of which intersect to determine what the mix is going to be of personal control versus perhaps carrier control. And then, of course, you've got also the Googles of this world coming into the marketplace, as well.

Lee: Well, when I listen to that, I hear two things. Carrier control, to me at least, means investment in networks. Now, we have a global, ubiquitous GSM network. I can be reached anywhere on one number, huge value. But at the same time, the application has remained static. As I think Norman said the other day, voice has pretty much followed the same paradigm the past 100 years and the huge disappointment, frustration and, to be honest, anger, has been the lack of innovation. And if the operators keep control of that access network, then the problem is, innovation will remain stifled. So do you feel that - I need to watch here knowing the consulting work you do -- can you comment on, as you called it, the middleman stiffling innovation there, the mobile operator?

Dean: Well, I think first is, you need to separate out the access from the core and the applications. I think that as long as there is sufficient competition in the access network, you will tend to get increasing levels of openness, and indeed this is what you're seeing now. In a lot of countries, there are now flat-rate data plans, you're seeing the opportunity to use 3.5G to LGs like HSPA, Wi-Fi, perhaps WiMAX in some countries, multiple operators; and there you tend to find that you will see access become more open. In some places, it will need a bit of a regulatory kick. And certainly there are enough places now that you can just use a mobile-operator service to get to whatever capabilities you want, whether on the web, in a corporate network or in the operator's own domain if they have a particularly good service.

Now, what you also then have though is openness on the device side and in the core and application layer. I'll talk about the device layer in a minute, but the core and application later, there is clearly a lack of innovation and sort of the proposed solution of IMS, if anything, exacerbates the problem in my view. It's very, very difficult for innovators to develop to IMS as a platform in the way that they do for the web. There are lots of interoperability issues; model of the standards are there. The idea of two men and a dog in a garage somewhere is more likely two men and a dog and a thirty-person legal department if you want to get a service adopted across multiple networks.

So I think that there's certainly some issues there. And to be fair, the carriers themselves are recognizing limitations. It's notable that IMS has had very little traction among cellular operators thus far, because they don't understand what the business model is. You know, frankly, voice works fine on circuit switched in most cases. And other things like IM and video-sharing are sort of almost prototype applications rather than revenue-generating ones.

So I think there is definitely a gap for innovation. Interestingly, some of the operators, the more forward-thinking ones, are looking to enable that by perhaps exposing their communication capabilities as APIs to third parties. I'm thinking about BT playing around with a thing called Web 21C and the Canadian operator Telus and others who are at the early stages of this. But it is early days.

It's also worth saying that on the device side, there is an interesting tousle around openness of devices. Some mobile phones are locked to a specific carrier. It varies hugely around the world, though. Roughly 50% of phones are sold through carrier channels or are carrier customized, and about 50% are sold separately. So if you go to parts of Asia, for example, you'll buy your SIM card from one shop and your handset from another. In North America and Japan and Korea and, to a lesser degree, Western Europe, that's historically not been the case, especially where handsets are subsidized, if you want to use that word, by the carrier. You could make an argument that if you're give a subsidized phone, it gives the person doing the subsidy, the operator, some sort of legal or moral authority to limit what you can do with it, whereas if they basically said, "Well, if you want to pay full price for the phone, then you can use it how you like; but if you want a free one, then you use it on our terms."

So I think we're going to see some interesting developments over the next 18 months here with a lot of carriers at least appearing to support greater levels of openness.

Lee: Okay. It reminds me of the comment that Martin Gaddis made about the Amazon book reader [Kindle], that it comes with connectivity and it's --

Dean: Yes.

Lee: -- pretty much an application tethered to connectivity and an application --

Dean: Hmm!

Lee: -- pretty much the same way voice and SMS are connected; you know, it comes with --

Dean: Yes.

Lee: -- connectivity and the apps. So if we subsidize a handset, the argument can be made to some degree that we can dictate what applications run. So I think you're correct in the use of the word "tousled" there.

As an end user, I want to see the explosion of innovation we saw in the '90s around the World Wide Web and the PC, and I want to see that end up in a handset, because a handset has far more opportunities because it's on you all the time, it's a personal device and all the other things that we know and I won't elaborate on.

Dean: Yeah. I agree in principle that there ought to be a mechanism for the type of viral adoption of cool new stuff that we see on the web. If you think about the first time you saw Google Earth, Facebook or whatever your favorite thing is. You probably on something like Google Earth, if you're like me, you probably wasted about three hours playing around with it and then emailed it to your friends, irrespective of what service provider they have, you emailed them "You need to get this now." And the thing is, there's no way you can do that in mobile at the moment.

You can't come across something cool and then tell your friends about it, because you think, "Well, what phone have they got? What network are they on? Are they on 3G, are they on 2G? Are they on CDMA, are they on GSM?" and so on. That mechanism for viral adoption isn't there. The mobile phone isn't the vector for the virus [sic: viral adoption] yet.

Now, that's partly down to the operators; but to be honest, it's also partly down to the nature of the handsets. There is huge fragmentation in handset operating systems. I've got a 3G phone where I've switched the 3G off on it to save battery, for example.

It's very difficult to predict what everyone else's capabilities are, and that makes it very difficult for innovators.

Lee: So it may be jumping ahead in what I wanted to cover with you. But are you able at this point to pass on any comment on the Open Handset Alliance's Android?

Dean: I have to say I haven't had as much information about how they're going to achieve what they want to achieve as I would've liked. I think to me at the moment, it's not much further beyond the stages of yet another platform in my eyes. Yes, I know it's Google, so you have to make allowances; but you essentially are dealing with a company which is run by idealists with lots of money, so they're in a position to make things happen. And it does seem to get support from a lot of people. But on the other hand, for that matter, so does Symbian and so does Linux and various other platforms for handsets, and fundamentally you can't escape the fact that 35% of the world's phones are based on plain, old Nokia Series 40 feature-phone platform. And it seems that large chunks of the world are quite happy with with very, very closed phones, not even like lock-down Smart phones, but basic phones which do voice, SMS and have an alarm clock.

Lee: (Laughing.) Dean, you always make me laugh. So

Dean: Yeah?

Lee: when you spoke, picking up on an earlier point you had that you touched across, can we go on the record and say, "Dean says that IMS is dead in the water"?

Dean: No, I don't think it is dead in the water. I think it's certainly not being deployed in the way that it was first envisaged by the 3GPP. It's deeply ironic that, in fact, the early implementations of IMS have not been mobile at all, but have been used for fixed voice over IP and also by some greenfield WiMAX operators. That's certainly not what 3GPP had in mind when they standardized it.

Bits of IMS are being deployed for certain things for certain operators. It makes sense in terms of consolidation of core networks. If you're an operator that's historically on multiple IP networks and you want to blend them all into one, it's an off-the-shelf blueprint about how to do it, and that may be easier than starting with a blank sheet of paper.

But there are certain elements of IMS which are not fully standardized or people haven't decided how to use them, say, particularly there's an element called the HSS, which is the subscriber data store, and at the moment that's not going anywhere very quickly. Conversely, some of the security functions and just some of the ways they sort of scale up session-based services, there are some upsides. But I certainly think the original IMS vision certainly has undergone an awful lot of rework.

So I wouldn't say it's dead in the water.

But it's sort of morphing into the more general web services, IP and IT space. I think if we come back in two or three years' time, there will be bits of network elements which are identifiably IMS elements; but I don't think we'll see too many people with a complete end-to-end IMS here and particularly not a standalone IMS.

You'll find that people are blending applications from a variety of sources. So essentially you've got Internet applications, you've got enterprise applications, IMS, you have maybe TV, old 2G and intelligent-network stuff which still works and various other sources, service/delivery platforms, and they're all having to be blended in some sort of network middleware with gateways here and sort of other interoperability platforms there. So I think we'll end up with a real mix at the end, to be honest, but not pure IMS.

Lee: You just said that the home subscriber server is not going anywhere, and yet it's the essential IMS component, would you not agree?

Dean: No, I wouldn't say it's an essential IMS component. The issue is that the HSS, no one is really sure what data you want to store in there, whether it's going to be centralized and distributed. If you think in some scenarios, if you were going to run every piece of subscriber data, you'd have things like people with mobile TV, there's useful information about how often they switch channels; but do you really want to create some huge data warehouse which has everything from your address and your billing details right through to just sort of whether you flip between adverts in the break on a mobile TV or sort of what the search terms are on the web or just too much subscriber data. And different applications want access to different subsets of it, and so it's almost impossible to try and standardize what would go in an HSS and what would go elsewhere.

Lee: Okay. I see how it's a huge obstacle personally to standardize what's in it, and from what I had heard you can't even put small amounts of data in it, because nobody would agree to extend it at all. People have been lobbying for that now because not even small amounts could be added in a custom fashion, too.

Dean: Well, it's one of those areas that's horribly political, and I think that's one of the reasons why, if you like, the full IMS is slow. That is one of them. There are some other issues around things like prepaid billing or online charging, it's called. There's issues about interoperability between IMS and non-IMS applications.

There's something that I've done a little research on around the lack of IMS handsets and the lack of IMS handset standards or specifications.

So as I said, there's pieces of the puzzle which are there and which are useful. I don't think the whole thing is going to disappear; but at the same time, the original vision of IMS doesn't appear to be realistic unless perhaps you're a greenfield operator or an operator which has such tight control over its own ecosystem that they can enforce whatever they happen to like without worrying too much about interoperability and standardization.

Lee: I'm going to ask you hopefully for a yes or a no answer to this.

Dean: (Laughing.)

Lee: Is IMS a telco savior in terms of ensuring continuing revenue far into the future?

Dean: No. I think that it's possibly a telco savior in terms of costs and cost management, though. If you look at the way that, say, BT positioned its business case for 21CN, its next-gen network, which isn't quite IMS, but it's sort of IMS-ish, certainly the way they've positioned it in the financial market is ... I can'tremember exact numbers, but roughly speaking it'll cost us £10 billion pounds and it'll save us £1 billion pounds a year in OpEx. Anything we get on top of that in new-services revenue is a bonus --

Lee: I have to add in, it doesn't use IMS.

Dean: Yeah. I'm paraphrasing Lee, that type of ... there is an argument that some form of IP NGN to carriers is essential to manage costs.

Lee: NGNs don't need any IMS standards.

Dean: They don't need any; but what I think you're seeing is that an increasing number of them are IMS-like. So you do have organizations like TISPAN, ETSI TISPAN, which is converging the 3GPP view of what constitutes an NGN. And it will be different in certain areas. But I still think that the underlying trend is towards some form of standardized IP architecture which is suitable for carrier business models. And on top of that, there will be also overlaid more Internet web services-like approaches to business.

Lee: Okay. So what I was supposed to be asking you about today was about wireless technologies. I did'nt intend to get stuck into the IMS thing; but as you know it's a major point, because it's really a hinge point, a lynchpin, actually, of the traditional telecoms industry.

Dean: Hmm.

Lee: So I've been meaning to ask you about the wireless technologies.

Dean: Yes.

Lee: You know about WiMAX, LTE, you know, Long-Term Evolution; Wi-Fi mesh, 4G, UWB, software-defined radio, High-Speed Data Packet Access and so on.

Dean: Yeah.

Lee: So would you just do a sweep of those technologies?

Dean: Right, okay. And this is going to be sort of a personal and Disruptive Analysis view of the world. Okay, WiMAX start off. WiMAX seems to go through waves roughly every three to four months of optimism and pessimism of the, oh, yes, it is; oh, no, it isn't.

My general view is, WiMAX will be important for non-phone devices, and so I'm thinking here of devices that might be PDA-type, they might be PCs, they might be games consoles; roughly speaking, things with large screens and large batteries.

The big question around WiMAX is around spectrum availability and how much spectrum and what type of spectrum it can use. In the U.S., you have 2.5 gigahertz. In various parts of the world you have 3.5 and maybe 2.5, which are great for fixed or outdoor uses, but have a problem and don't go through walls very well. So either need to have some sort of indoor-coverage solution like a femtocell or revert to Wi-Fi indoors, and that's an issue.

LTE, Long-Term Evolution, seems to be coming out as the medium- to long-term winner in this sort of war in what we generally call the pseudo-4G stakes. At the moment, there are no official 4G technologies, because it hasn't been defined yet; but there's lots of both marketing 4G and also what might be termed real candidate technologies for 4G, to be considered for 4G status. An LTE is probably the frontrunner there.

There's a lot of tests ongoing. LTE in theory offers extremely high speeds and efficiencies coupled with mobile operators working on a thing called NGMN, Next Generation Mobile Network. They appear to be trying to sort some of the ecosystem problems upfront so that both the network and the handsets and the transport infrastructure are all being developed at the same time. I'm expecting it to start appearing commercially in maybe one or two things at the end of 2010 with bigger pickup in 2012, 2013. It's not going to be easy, though, and I think there's going to be a lot of challenges that crop up in the early sort of trials and test beds that need to be fixed.

Wi-Fi mesh, I've been a deep skeptic of Metro Wi-Fi, in particular, for the last couple of years. I just don't see it as a useful medium for anything other than fairly boring applications like municipal collection of traffic-warden data or maybe CTTV cameras.

There will be some uses for it, but I don't think it's going to be around person-to-person communications. You've got a big problem with Wi-Fi meshes that Wi-Fi meshes tend to be outdoors and, once again, you have an issue of the outdoor network and the indoor network don't play together very nicely.

4G, as I said, I mean, at the moment there's lot of things using the term 4G; no one actually has precisely defined what it constitutes, but roughly speaking, I mean, anything that crops up from 2012 onwards, really.

I think there's a lot of challenges with the radio network, again on spectrum issues and these handset issues that need to be resolved. And I think initially it will be data applications, although interestingly you also run into the issue that any voice on 4G has to be Voice over IP because it's expected that they will be all IP networks.

UWB, Ultra-Wireless Broadband if I remember the acronym correctly, is one of those things that people talked about for a while. My understanding is that it's most likely manifestation is going to be in short-range communications, a sort of very far Bluetooth type use case.

At the moment, it seems to be going very safely. It's not something that people are talking to me about very much at all. I can see it having relevance perhaps in the home for distributing video signals around the home and, as I said, this sort of more familiar Bluetooth-type usage caters or sharing data between devices. I don't see it as a wide-era technology.

Software-defined radio is one of those things which needs to be defined itself rather more carefully. In theory, you have a chip on a handset or in the network which can reconfigure itself to deal with multiple protocols, multiple technologies, multiple bands; but it's not as easy as that.

Firstly, that tends to use a lot of power. Secondly, you have a whole range of issues to do with regulation, tests and measurements and sort of what are the protocols by which the software defines new radio types. How can you be sure that a given radio type is permissible in a given country, or is it licensed or unlicensed spectrum? You see SDR used in military radios, and I suspect you'll start seeing it in the mobile infrastructure first where power is not so much of an issue. I have my doubts whether it will appear in its full inclination on handsets for quite some time, although you might see certain components become a bit more configurable perhaps with the radio side of the handset first, where you've got maybe filters which can adjust their characteristics or amplifiers.

3.5G, whether it's sort of HSPA or the CDMA/EVDO, Rev A and beyond, they're the real world now. There is a rapid growth in the uses of 3.5G technologies, particularly for PC connectivity at the moment. There's a huge ramp-up of sales of things like 3G modems, either in some cases built into PCs, but increasingly just sold as a standalone HSDPA or HSUPA dongle USB modem. I know that they're growing incredibly quickly at the moment. And they're quite interesting, because to some extent your HSDPA is, if you like, the bug fix for 3G; it makes 3G work properly. You get reasonably good download speeds, you get quite decently low latency, depending on the spectrum used you get okay coverage. It's generally made 3G into a more day-to-day useable technology, and HSUPA and HSPA Plus will improve that still further.

As I say, the moment is mostly focused on PCs. Although there are a number of HSDPA handsets, even a handful of HSUPA handsets, they're pretty few and far between at the moment particularly in terms of being used in anger for anything other than occasional web browsing. I'm expecting this to change over the next few years as it starts to become more feasible and more desirable to run the Voice over IP over the 3G network. It's possible already, and there's a few operators who are playing around with it, as well, as well as a lot of separate standalone challenger software providers, things like Truphone and Skype and Fring. And I think that running VoIP over the 3G network becomes a lot better in terms of quality and in terms of spectrum efficiency with 3.5G and beyond.

Lee: Dean, for the purposes of those listening out-with the UK, because you're in the UK.

Dean: Yeah.

Lee: when you spoke about the Hutchison 3G and dongle there, how much do you pay a month for that?

Dean: I'm paying £15 pounds (~30 dollars) a month, including tax, for 3 gigabytes of data on a 12- month contract with a free modem.

Lee: Okay. And what speed, are you getting?

Dean: Well, it varies; depends where I am. Most of the time it's a couple hundred kilobits a second; but it gets up to about 800, 900 on good conditions. It's sufficient, I would say. Sometimes it drops down, particularly when I'm at my local Starbucks as a basement and it may vary, I'm lucky if I'm getting much more than 100 kilobits. But frankly, it is useable for most applications, and when I'm in good signal coverage, it's fine for, say, running Skype over.

Lee: Okay. And one of the HSDPA handsets I think everybody will know, a lot of people that have not heard of HSDPA --

is the Nokia N85, correct?

Lee: So, for example, I've got UK data SIM, as well, which I use when I am back there, and I pay 12 GBP a month, which is 24 US dollars a month; it's unlimited. And I actually get 1.8 meg in quite a few of the cities.

Dean: Well, that's good.

Lee: That means I can access the Google apps on my mobile phone, place voice calls over 3G free of charge around the world using Truphone
(00:30:07)
Dean: Hmm.

Lee: and I can also bluetooth it to the laptop for Internet access. So things are moving. But you had mentioned to me -- I may be going off topic; but you just informed me about your new roaming deal. Do you want to just mention that?

Dean: Yeah. I was going to say, actually, one of the things that Hutchison 3 does which is very unusual and which has caused quite a bit of consternation in the industry is, it doesn't charge a premium for data roaming. So I can use that modem when I want another Three [Hutchinson] network, and they have them in Denmark, Sweden, Italy, Austria, Hong Kong and Australia, without paying a roaming premium. And I think that that's a very interesting approach and one which I suspect doesn't sit very easily with a lot of the operators who are reliant on roaming charges both with voice and data as a source of margin.

Conversely, I've got a T-Mobile account, as well, and that charges me £7.50 GBP; that's $15 dollar per megabyte when I'm roaming, even when I'm going from the UK to Germany, their sort of home country. I imagine they have a huge fiber between the UK and Germany, but it still costs a huge amount. It's vastly cheaper to buy a satellite modem.

Lee: Okay, because I'd been speaking to Brough [Brough Turner] the other day and we'd been speaking about rip-off data-roaming rates harming the industry, and it's not just harming the industry. What we're doing is, we're blocking innovation, because you just can't stay on the Internet via your cell phone and be mobile at the same time.

Dean: There are some improvement in data roaming; but frankly, there's still one or two zeroes too much in there.

Lee: I wonder why we even have the concept of data roaming.

Dean: That's a very good question. A lot of it has to do with the actual structure of the cellular network where my understanding is that most of the data traffic will generally be back-hauled through to the home operator's infrastructure rather than breaking out onto the Internet locally. And I think the assumption there is, when it was designed was you would be using the operator's or your home operator's services rather than the real Internet. If you're just accessing the web, there's no reason why you shouldn't be able to sort of break out locally in whatever country you're in subject to maybe some content filtering or something, I guess.

Lee: So you've hit upon a key issue. Years ago, I had been arguing quite furiously about the whole concept of home-network operator's a very bad idea for everybody and that basically people want to build networks, well, they should have beacons offering you connectivity, and your device should be able to scan these, offer you connectivity options at set prices and you connect locally. This whole notion about back-hauling you off to some other country just because you took out a SIM card there I find quite appalling, but do you see any change in that direction?

Dean: Well, I suppose the largest change is going Wi-Fi, which has sort of proven that concept to a degree, perhaps not always with a greatest deal of user-friendliness, but you can and do use Wi-Fi on a PC from whoever the local providers are. I've also seen some movement ... one of the nice things about having the USB modems for PCs rather than the embedded 3G modems, which theoretically seem more elegant, is if you want to, you can buy another local one [SIM]. I know in some countries you can actually rent them for the duration of your stay, paying local prices.

Lee: Not exactly seamless.

Dean: It's not exactly seamless, no. But it is certainly some way towards that model. You may find that with WiMAX, you are also more able to. But the problem is, a lot of this would be contingent on the connection manager's software than your PC or in your phone, and that's an area which is lacking standards and it's also lacking at the moment the motivation of people to standardize. When you consider that a lot of particularly the phones are sell-through mobile operators, they're hardly likely to specify a software layer that allows you to use one of their local competitors in another country.

Lee: Okay. Do you think long term, we're going to can the notion of a home operator? If we can the notion of a home operator, that is huge structural, fundamental change within the industry. Do you see this happening long term?
(00:34:55)
Dean: Not easily. Maybe for some users. I mean, you've got to bear in mind that the majority of users don't roam that often. There is a slight difference in, say, North America, where you roam nationally outside of your operator's coverage areas. But we're talking about international roaming. Apart from business travelers and a few other frequent fliers, it's one of those annoyances which doesn't cause enough pain for the mass of consumers, and the pain it does occurs once or twice a year on holiday, frankly the best way to resolve that is probably through regulation rather than necessarily needing them to sort of select different operators each time they get off the plane. There's also a whole Lee of issues around numbering. It's fine if everyone uses a Skype ID or an email address or email number or whatever. But as long as we've got ordinary telephone numbers, there's a whole range of issues with how you get inbound calls.

Lee: Okay. So can you pass comment on let me begin this question again Dean. At the moment I'm playing with the Truphone who are going to be speaking [at eComm 2008], and I can pick up my cell phone and place a call to something like 1400 free of charge.

on my Nokia 85, I'll speak to the wireless-access point in the house and it gives me a very good quality call and I don't pay a thing and I'm calling landlines largely in the United States from Austria.

Lee: And when I'm in the UK when I'm using my data-roaming SIM there, I can just launch a call over 3G and be calling again to the States free of charge on my cell phone.

Lee: So how do you see these third-party Voice over IP over 3.5G players?

Dean: I see them becoming increasingly important both on a standalone basis and in partnership with the traditional carriers. As I said before, the traditional carriers ultimately have to go to Voice over IP if they want to do 4G; the question is how they get there and when they get there. And I would argue that one of the ways for them to gain experience of full mobile Voice over IP is to partner with people who are doing it already. There's a whole sort of range of reasons why that is.

Lee: Why do we care about a mobile IP?

Dean: Well, as I say, what do we care? You shouldn't necessarily as an end user need to care. The operators, as I say, firstly, they're going to be forced into it ultimately by 4G anyway.

Lee: But why do they have to go to 4G?

Dean: That's assuming ultimately we want to have higher and higher bandwidths and better capacity-utilization, spectrum is a finite resource. So if we want to have in theory mobile broadband and ever-faster mobile broadband, then we have to look at everywhere we can to optimize the spectrum we have access to. One way of doing it is using IP by using some of the essentially the sort of multiplexing characteristics and pushing intelligence down further into the radio network. Certainly once you get to a certain point, it is possible to get more voice calls per megahertz per cell with Voice over IP than it is with today's circuit switching.

Lee: Okay, because when I was at Hutchison, the bandwidth usage of VoIP was much greater than TDM, the bytes on the wire.

Dean: Yes, that has been the case. But certainly you have issues like packet overhead, which is true on the initial 3G networks. As you get to later generations of HSPA and then LTE, there is things like robust header compression, better packet scheduling and a range of other techniques in terms of, say, the receivers, which actually improves the spectrum efficiency above that of circuit switching. Sort of the crossover point comes around about HSPA with all of the sort of optional radio tweaks turned on, and then from then on there's a definite benefit.

Lee: Are we limited in processors and handsets to achieve that?

Dean: Yes, and by extension also power; however, one of the things there is the ability to optimize that, and it hasn't been explored. There's been a lot of activity over the last three years on optimizing Voice over IP over Wi-Fi on phones; but we're certainly not there on looking at the cellular side of packet voice yet. And one of the things that I'm aiming to perhaps try and capitalize is some more appreciation of the fact that that work needs to be done, and ideally it needs to be work done before the mobile industry is forced into it by LTE.
(00:40:28)
Lee: Okay. So I cut you off on commenting on third-party VoIP over 3G ...
the likes of Truphone.

Dean: Yeah. So I think that there's sort of certainly three or four companies that are well-known in that space: Truphone, Fring, Skype, some others like Challenger are doing it and a few more, as well.

Lee: But Skype is only on Windows Mobile, and Windows Mobile, am I allows saying, sucks very, very hard.

Dean: That's coming back to what I was saying before about --

Lee: I'm talking about my mobile.

Dean: Yes, but there's mobility, nomadicity. It depends on the use case. The way I see communications generally is, people are fragmenting the ways in which they communicate with other people, depending on their precise circumstances and preferences.

And so there is a valid-use case, which is if I am a mobile worker and I have a one-hour conference call where I want to call into a U.S. number, then I may well do that from a PC. And if that's the case, whether I'm using 3G or whether I'm using Wi-Fi, there is a usable ... actually, I prefer the PC for that type of call, because it means that, for example, I can look at a presentation on the screen, I can look at the website of the client, the customer or vendor that I'm talking to. So in those circumstances, you might well use a VoIP client on a PC by whatever bearer.

Conversely, for handsets, one of the things that I'm thinking about is whether you might want to blend Voice over IP with other applications. We're very early days on this, but certainly the general concept of mobile VoIP mashups is one that has been sort of mucking around quite a lot recently. At one level you can have voice inside a game; but it could also be social-networking oriented, it could be encrypted voice, it could be mobile voice as part of a corporate application. And so I think there's going to be some interesting innovation around that area over the next few years. I certainly don't expect it to suddenly account for a big chunk of the total number of voice minutes from a handset; but it could become an interesting share of the voice-over value, however you want to measure that.

Lee: As you say there, it may not end up being huge. Certainly in short term in terms of minutes, but you may find others enabling applications ...

Dean: Yes.

Lee: I don't want to use the word "converge", but including voice in them --

Dean: Yes.

Lee: -- which make a lot of value in those minutes.

Dean: Yeah, yeah. And at the moment, we're still really exploring those. It could be consumer applications; it could even be things like recording voice calls and using a phone as a dictaphone which might use this; it could be advertising-related. It could be customer service; it could be driven by business processes and enterprise applications. Now, in some of these cases, you could do things by circuit-switched, for instance, as well, if you had the appropriate APIs. But there are some things that will evolve that will be VoIP only. One very clever concept that someone suggested was around from a gaming point of view using stereo cues to locate people and so that you could potentially use Voice over IP with sort of all the other positioning technologies so you could hear your teammates behind you and to the right if you were wearing a headset, let's say.

Lee: Okay. Can you pass comment on 4G, what is it, when is it happening?

Dean: At the moment, what is it? At the moment, 4G is a marketing term used by all and sundry for different things; however, officially 4G will be a set of technologies defined and agreed by the ITU, the International Telecoms Union.

My understanding is that we're currently going through -- I haven't looked into the sort of the full regulatory process here -- is that we're going to go through a set of candidate technologies being proposed to be 4G based on some criteria. I'd imagine those are going to be LTE, perhaps the next version of WiMAX, which is 802.16M, I believe, and possibly CDMA/UMB, as well as maybe one or two others; there's possibly going to be a Chinese one and, you know, whatever else might come out of the woodwork. Now, that in theory will be sorted out, I guess, over the next year in terms of defining what is a 4G technology and what is not.

Then we have a whole range of issues around allocating spectrum for them, getting test technologies up and running. Rollout of networks, as I say, you might see a couple at the end of 2010; but I think they're sort of the big, round ramp up comes in 2012, 2013, 2014.

Lee: So our industry may have spent huge sums of money on 3G licenses, only to find itself quickly being run into 4G.

Dean: It sort of is; but one of the things I think you'll see, that 3G and 4G are starting to blend. Certainly from a spectrum point of view, you probably won't have dedicated 4G spectrum; you will have spectrum that's usable for 3- or for 4G. So I don't think there's going to be as definitive a cutoff as we saw with 2G/3G. I think people are going to be more migrating rather than trying to change everything overnight.

Lee: Who gets to play in the 4G game, then? Who gets access to that 4G spectrum?

Dean: Anyone with lots of money, I guess.

Lee: Will we see auctions again?

Dean: Yes, absolutely. I mean, I see auctions; I see the existing 3G operators being able to reuse their spectrum for 4G if they want. You're moving a lot of territories towards what's called spectrum neutrality or technology neutrality, where as a service writer you basically bid for spectrum and then within the constraints of what you're allowed to do in terms of interference and so on do what you like with it. If you want to deploy a 3G technology. oh, great. Want to deploy a 4G? Fine. And then you have to make a decision about the business models, about whether you can get multiple-mode devices which can cope with all of these, whether you've got a business case for the 3G only or 4G only, whether you want to do backwards compatibility to 2G, whether you want to do hybrids with Wi-Fi and WiMAX. It's going to be messy, and there's probably not going to be a single unifying theme there.

There's also a whole range of issues around the radio network, which needs to be looked into. I mean, particularly around the performance of things like beam forming and MIMO, which in theory are key elements of all the 4G radio technologies, but actually in reality are pretty poorly proven as a sort of scalable, mass-market, wide-area technologies.

I think one of the things that's going to crop up over the next ten years, and see if you can realize this, the RF stuff isn't as easy as they think, either because of interference or because of power issues.

Lee: Who thinks, Dean?

Dean: Well, a lot of the application providers and some of the network operators just think that we're going to end up with continually scaling mobile bandwidth, for example, and mobile capacity and that the phones will follow and everything will be good; it'll be like home broadband on the fixed network where you go from ADSL, to ADSL2, to ADSL2+ to VDSL. It's not as easy as that. And I was talking to a test vendor yesterday who is sort of raising issues around the performance of individual brands of handset, for example, HSDPA; and in fact, he only needs one like rogue handset with a poorly performing chip set to actually affect everyone else in that cell because of the way the radio networks are being designed. So it's one of those areas that's not as easy as it looks.

Lee: Okay. Now, you touched on open spectrum there, and I haven't had time to look into it because I've been so busy organizing this conference. But I gathered that the OFCOM, the UK regulator, was following the FCC in terms of planning to allocate open space; and if this is true, it also seemed that OFCOM was at odds with the European Union, who were more on fixing a spectrum to a particular application.

Dean: The European Union is also going down that route. OFCOM is one of the louder advocates of technology neutrality. There is a European Commission thing called WAPECS, which is around wireless-access policy, and that's a heavily contentious issue on how technology neutrality is brought in. I think it's one of those things that you were going to see, the regulation, legislation evolve over the next couple of years. Let's say the UK is one of the greater advocates of flexibility.

On the other hand, it's worth saying that I think some of the aspects haven't been fully explored, particularly a lot of the standards, say, some of the 3GPP-defined bands for things like UMTS have been defined and the conformance tests have been defined in the expectation of single-application spectrum. I don't think it's necessarily the case that there's been as much consideration given to the testing and conformance as standardization aspects, for one thing. I think it's possibly some of the standards are going to have to be gone over again and looked at to see whether the assumptions are still valid in the case of technology neutrality.

Lee: Okay. So I also heard and, again, I just haven't had the time to look at this, do you know if Google had been aiming at acquiring, or at least rumors of it, UK analogue TV spectrum when it becomes available over the next few years.

Dean: Well, I mean, I would imagine that Google, given its currently stated interest in the U.S. 700-megahertz options, I'm sure is looking at spectrum available elsewhere. It would be strange for Google to have a U.S.-only plan for spectrum, I think, even though the U.S. is obviously sort it's home market. So I'd certainly imagine that it is, particularly because there's also European Commission looking at whether it will be possible to run non-TV services in analog-TV spectrum using what's being referred to as the digital dividend. And certainly that's an area that the European Commission and the European regulatory communications agencies have been looking at.

Lee: Okay. So have you specifically heard these Google rumors towards the UK market?

Dean: No.

Lee: You haven't heard this.

Dean: No, I haven't heard this specifically; but it doesn't surprise me.

Lee: Okay.

Dean: Yeah, particularly because the [UK] TV spectrum is in the same general band as the U.S. 700 megahertz; and I can't remember the exact band that the analog TV is in, but it's something like 470 to 860 megahertz. So clearly, if Google was ... well, let's hypothesize that Google wins the spectrum in North America for 700 megahertz. If it could essentially reuse the same products in other markets, which also have approximately 700 megahertz frequencies, then clearly they'd want to look at that very hard.

Lee: The EU seems rather fragmented on this approach to spectrum allocations, so I wonder how easy it would be for them to move beyond the UK and acquire spectrum in the other European countries.

Dean: It would be patchy. Again, there are attempts in Europe to do a sort of a harmonization-lite is probably it; so it might be harmonization with exceptions, or it might be there are certain things which are optional, but the options are fixed. I don't think you're going to have the same situation as you do with GSM, where it's like 900 megahertz across the whole of Europe and 1800 [megahertz]. So I think it would be less clear-cut than that, but not the total free-for-all.

Lee: Okay. And I don't know too much on RF side, so do you mind me asking about the 700 megahertz and that particular frequency range. Is that rather low in terms of range?

Dean: Actually, in terms of range it's very good. Roughly speaking, the lower the frequency you get better range, you need lower power. It goes through walls better, but you get lower capacity.

Lee: Yeah, I had it backwards. So how limiting do you feel the bandwidth is there?

Dean: Well, it depends on how it's deployed. I mean, in theory you could have one 700-megahertz cell tower covering a huge area; but then it would be sharing comparatively small amounts of the spectrum for everyone in that area. Now, the other way you do it is, you try and combine it cleverly; so maybe you use your lower frequencies like 700 for rural areas, and you use higher frequencies in places where you've got more cell sites in town. Now, that then depends on whether you're starting from the position of having lots of cell sites like the incumbents or whether you're a new entrant like Google. So it's a very complicated set of criteria to try and optimize for. You also have for any given location, it's easier or cheaper or more expensive or more difficult to get hold of new cell sites. There may be regulations against it, or you may have problems with getting power and back-haul there. And then also you also have the possibility of doing all of this indoors using things like femtocells.

Lee: Okay. And we should wrap this up, so I wanted to drive towards this question. When can garage-based hackers -- you know, two men in a garage and a dog, as you put it -- put their attention on the radio side and get out with this, having their hands tied by mobile operators in order to actually innovate?

Dean: You could argue that the application there, you already can. If you're prepared to port across multiple different handset operating systems. In terms of when can you actually hack the radio network, I think that that will be a very long time coming in terms of you coming up with your own radio protocols and being able to play around with them. There is a finite amount of radio spectrum, and there is a risk of creating what's called the tragedy of the commons, where you let everyone do whatever they like and you create so much interference that everyone ends up with nothing. And so I don't think that we're gonna see regulators get out of the way on spectrum, possibly ever, although there are possibly grounds for having certain defined parts of spectrum as playgrounds for people to experiment with. But I certainly can't see the regulators sort of basically turning off regulation on 900 megahertz or the sort of main transmitters, not just for the communications reasons, because all the other stuff like military and radar and emergency services that need to be protected.

Lee: Leaving aside people being able to play about how they wish, how soon do you see people being able to engage with open spectrum in terms of, say, the likes of Google if it wins spectrum? Do you see this coming along more in the UK, more of this where the frequency has been won by somebody, but that somebody's committed towards making that open towards whatever they wish to achieve and let end users actually play on a piece of spectrum that they won?

Dean: I certainly see wholesale models becoming more prevalent, where you have maybe Google or someone else who builds out an infrastructure and then comes up with ways of wholesaling it to other service providers. I don't think that you would necessarily be able to do that down to the individual or developer level for a very long time. I mean, I think it could be quite difficult to have a sort of Google API for the radio network, where you essentially define your own capacity because of the need for physical infrastructure like cell sites. You know, there are a lot of constraints, both technically and commercially, for whoever is the spectrum operator.

Lee: Dean, can you tell us what is meant by open spectrum and what the opportunities are?

Dean: I have to say, it's not a term that I regularly hear. There are so many uses and misuses of the word "openness" in mobile at the moment. But I suspect it could be interpreted in a lot of different ways. I mean, what's your argument for that? What's your interpretation of "open spectrum"? What do you have in mind?

Lee: That you're free to experiment with applications at that frequency rather than be bound and committed legally to a given application.

Dean: Okay, I'll go another stage further down the rabbit hole. How are you defining "application", because, I mean, there are so many --

Lee: Voice.

Dean: Well, you already can. I mean, there's nothing stopping you doing that. I mean, we've already talked about open access or 2- or 3G networks. With an open device on essentially a flat-rate data plan, you can run whatever application you like. You're a little bit constrained on the quality of service, the coverage of whatever service provider you're using. But at application level within those constraints, you're free to play around. There are some operators that have restricted terms of service which say you can't do VoIP or you can't do Bit Torrent, and they may have packet inspection gear. But that's certainly not all of them, and within reason if you've got an application which runs over the top, whether it's social networking or whether it's ERP, you can run that.

Lee: Dean, are you able to pass comment on open spectrum defined as unlicensed spectrum that's for use by all, a commons type model?

Dean: Right. So there, you're talking about the equivalent of, say, 2.4 gigahertz, what used to be called the industrial, scientific and medical FM and there's a couple of other bands, as well, which is where Wi-Fi is, where as long as you conform to certain regulated characteristics in terms of sort of total power outputs and emissions into other bands, you can pretty much do what you like. Clearly, Wi-Fi in particular and various other models anything from garage-door openers to microwaves ovena have proven that there is value there. What there isn't at the moment is any particular concerted lobbying or regulatory effort to get more of that. It was interesting at the World Radio Congress at the end of 2007, there was very little noise that I heard about getting greater level of spectrum allocated for that type of purpose. I don't think it's been adequately proven that, say, the 2.4 band is too congested in most places or that other bands like 900 megahertz in the U.S. or I think various things in the 5-gigahertz range are now unusable. There doesn't appear to be any obvious political or regulatory motivation for extending it. I've asked a number of vendors and regulators before whether they liked to see more unlicensed spectrum, and the answer generally comes back, "We think there's enough."

Lee: Okay, Dean. Thank you very much for your time. I very much looking forward to seeing you next March in Mountain View at the Computer History Museum for the eComm, the Emerging Communications Conference. So again, thank you, Dean, and I will see you there.