Monday, July 30, 2007
I had a chat with the guy in the shop and he said that they were selling them primarily together with Orange contracts, usually with either just 4MB or 10MB of 3G data per month, although he seemed to think this was ample for most peoples' use of VoIP over the wide area. He said that the T-Mobile T's & C's on Web'n'Walk prohibited VoIP, although one of his friends had been happily using Skype over cellular with no problems.
What wasn't immediately obvious was where in the process the fring client was being installed on the phones - it certainly didn't seem to be operator-specified, so I imagine it could just be the retailer (not a big chain) gets them preloaded from a distributor.
On a separate note - I see quite a few E65's in circulation for general consumers, rather than enterprise customers. This particular one was in a very non-corporate shade of crimson red.
Tuesday, July 24, 2007
The first and most important takeout is what applications and customer types they're aiming at initially. It is absolutely evident that xMax is not aimed at service providers looking to compete head-to-head with dense, large-scale, telco-grade cellular networks with all the complicated machinery they require. Maybe on the far horizon if all goes fantastically well, but not right now or realistically within a 3-4 year view.
The target segment of operators is much more akin to a those that are happy with lowest-cost, best-efforts, ISP/VoIP provider approach. The base stations & infrastructure are aimed at competitive service providers wanting to create a bargain-basement, 'alternative' network that avoids the need to tackle complex wholesale and interconnect arrangements with incumbents (wired or wireless) except via Internet VoIP.
Now, there are quite a lot of service providers that would be happy to take the compromises involved - proprietary technology, no "name brand" handsets, little spectrum management, limited end-to-end QoS, no international roaming and so on. These are the same type of organisations that might also look at low-end WiMAX installations, or even the larger WiFi mesh approach. But it is important to stress that these are essentially wireless ISPs that also have a VoIP proposition - similar in many ways to WiMAX Telecom in Europe or niche UMTS-TDD players like Woosh in New Zealand. The difference is that those operators are starting with PCs (which can support VoIP) rather than phones.
All this is well and good - and may even enable 100's of thousands of subscribers in the US, where there's still a few million subscribers on regional cellular networks outside the 'Big 4' carriers. Many of these customers have limited service capabilities - even having to pay extra if they roam nationally out of their local area. But it's not going to quickly scale to millions of users per country for a variety of reasons.
The international story was also interesting, for example talking about its evaluation with Telefonica's Mexican arm. The angle there is to allow large players to enter into commercially challenging new markets at a low cost, not to replace their existing incumbent networks such as Telefonica's Spanish busines or O2. Fair enough as an entry strategy where it's hard (politically or regulatorily) to deal with incumbents, but again it's a bit of a lower-grade solution rather than a fully bulletproof one.
Less clear was the situation with other international markets outside North America, particularly with regard to spectrum. The company's position is that it expects its customers to have access to their own spectrum - but they were very vague about which bands were ideal, how much spectrum might be needed for a given application, or critically, how long it might take to support a new frequency band in the base stations or phones. I also found it slightly worrying that the execs were unclear whether necessary spectrum had to be designated for TDD or FDD use.
I specifically didn't go down the route of asking the company about the recent public spat with a Qualcomm employee (Phil Karn), who has aggressively been trying to 'debunk' some of the company's more high-profile claims for xMax. Nevertheless they proactively raised the issue & pondered on his motives. The execs didn't brandish statements about 'orders of magnitude' better performance but were keen to stress power efficiency. Given that nobody with an engineering background was in the meeting it wasn't a topic I wanted to focus on. Either way, I don't see worrying about xMax as exactly being #1 on Qualcomm's to-do list at the moment given its other predicaments, so I don't see Karn's mission as some weird conspiracy: I suspect he just likes an argument, although in my view there are better ways to win one than the route he's taking.
However.... the CEO did give me useful numbers which I hadn't heard before. Some other engineers may be able to get their teeth into these & yield some analysis.
Apparently, the 1st generation products should yield an aggregate of 17Mbit/s in 26MHz of spectrum in the US ISM band at 902-928MHz, after accounting for guard bands and overhead. Ultimately the company is aiming for 2 bits / Hz in this band, in what they refer to as a "4G" network. They are also intending to go for four sectors per site, yielding a total of 17x4=68MBit/s , or in later versions maybe 2x26x4=208Mbit/s
[edit - that's my calculation, and based on comments below I should have asked about spectrum reuse & probably a bunch of other stuff. Treat it as an upper bound, I guess]
If I'm reading the tables here correctly, that's good but not exceptional in terms of spectral efficiency. My engineering knowledge doesn't permit me to say anything about power consumption, though. Someone else may have a view on this. It's worth noting that range is a bit of a non-issue here as it simply spreads the given (fixed) capacity over a larger area. Again, possibly useful for a startup operator wanting to build up a best -effort network over time by adding base stations, but again that's not a viable proposition for a mainstream operator wanting to launch a large network on Day 1 without having to go back & acquire more lots cell sites subsequently while the network is already running.
There's a bunch of other issues I haven't touched on here that are also important especially around backhaul, standards, threats from US 700MHz services and regional exclusivity.
But my current bottom line is that it sounds like a solution for the "Wireless ISP with VoIP" market, but not a near- or mid-term threat to conventional cellular voice or data services. There may be some extra-clever stuff in how it does power management, but the rest of the solution is not 'telco grade', and it would take a long time to get there as there's no 'ecosystem' in place and no clear strategy to build one. Although they ultimately want to sell silicon, it's not obvious how (for example) Nokia might engage with the company to integrate xMax into its phones.
I've seen some recent comparisons between xMax and 3GPP LTE and major 802.16e WiMAX rollouts like Sprint's, and I think they're largely unfounded & unrealistic - they're completely different propositions aimed at different customer groups & usage cases. To my mind, xMax is more of a potential threat to low-end ISP-type WiMAX deployments, CDMA450 in rural areas, and the largest WiFi mesh implementations, especially in North America or anywhere else with 900MHz unlicenced spectrum. I'm unconvinced they've progressed much towards supporting other frequencies - they certainly don't seem to have any other obvious preferred 'profiles' for 2.6GHz or 700MHz or 2.3GHz or whatever, and given work I've done recently it would probably take a lot of time & effort to support even one of these to a reasonable degree.
(As with my previous posts on xMax, I'm going to be ruthless with comment moderation/deletion)
I've been checking it on mobile devices since 2000 - initially on a browser on a Palm PDA, hooked up via infrared to a Siemens SL45i phone connected over GSM. More usually, I use either a handset's WAP browser, or a normal HTML browser on a smartphone.
Recently, Yahoo has messed up, though. A lot.
Firstly I played with it's Yahoo Go software on the trial N73 I had running on 3's X-series service. It was utterly, utterly dreadful, the single worst piece of mobile software I'd ever used. Slow, temperamental, didn't have a 'refresh' button so I couldn't see new messages immediately, limited memory. Just ugh. Worse than the 2000-era Palm/Siemens/infrared user experience. And much worse than just using a good HTML browser set for full web pages, downloaded over 3G on a Nokia E60 I was also using at the same time.
And now it seems to be trying even harder to mess up. I've got a Windows Mobile HTC device that I use for web browsing on a T-Mobile Web'n'Walk plan. Historically I've done 'pull email', checking Yahoo a few times a day. Worked fine. But now they've gone an 'upgraded' (!) the site to its new Mail Beta. Yes it's a bit prettier, but takes longer to load in, and stupidly only displays 5 messages per page, with no obvious option to let the user change it to 10/20/50 etc. Hello? Yahoo? Tell me - what's quicker, scrolling down... or going back for another roundtrip to the network? Let me see, outside of 3G coverage on T-Mo's GPRS, I reckon that's 0.3 seconds for option 1, versus 7+ seconds for option 2. And at least up until last week you still had Mail (classic) as an option on the main Y! menu, but you've now removed it. Put it back, NOW.
UPDATE: I've noticed something even worse about Yahoo's mobile beta email. It only loads in a single page of text from messages, requiring the user to click 'more' after about 10-15 lines depending on your screen size. This is outrageously stupid, and obviously straight out of some 2003-era book on designing a mobile user experience. Guys: phones have scrolling functions! People use them regularly, eg for scrolling through their contacts lists or message folders. This isn't rocket science.
Yahoo - I'm a (very) loyal user, so please, please take your mobile mail strategy back to the drawing board before I have to switch to someone else.
Monday, July 23, 2007
Yes, I know that Google's talking about pitching $4.6bn for the US 700MHz spectrum... and yes I know that there's a 700MHz standard for UMTS going through 3GPP at the moment. But I'd have thought that femtos at that sort of frequency was fairly pointless, as the big attraction of 700MHz is that it's got great range & goes through walls easily.
The real advantage of 3G femtos in my view lies in 2100MHz 3G spectrum (ie most of the world today outside the US), and probably in the future in 2600MHz band. It's conceivable that Google might want to start bidding for those chunks of spectrum around the world, but I'm unconvinced that it's going to follow the classic cellular path (ie 3GPP UMTS or LTE) rather than something more Internet-like.
There's an outside possibility that Google might in fact want to do something with WiMAX - but at present Ubiquisys doesn't do WiMAX femtos, although chipset supplier PicoChip is certainly doing suitable silicon. Maybe that's what the investment's for....
Wednesday, July 18, 2007
Maybe someone in Brussels has had a sudden attack of commonsense and realised that best standard to follow for mobile TV is not to bother in the first place. There's no massmarket demand, no obvious business model, and lots of better things to do with the spectrum. Sure, there are probably a few specific niches for mobile TV to be successful, but the idea that 10's or 100's of millions of people in Europe are going to be subscribing to them and watching avidly is pure fantasy.
It seems that it does indeed come in both "operator-oriented" and "non-operator/enterprise-owned" variants.
The device's specific website for the UK has this to say about the WiFi functionality, which makes it seem like purely a service provider play:
If you have signed up for a qualifying BlackBerry® data plan¹ through your service provider, you can also use your BlackBerry® 8820 smartphone to access your BlackBerry data services over virtually any Wi-Fi network.
Voice Support over UMA/GAN
If your service provider offers UMA/GAN support, you can also take advantage of full voice functionality² over any Wi-Fi network supported by your service provider
OK, so the much-rumoured UMA support is embedded, no particular surprise there. I guess the major use case will be so that employees on T-Mobile US can use the thing at home where coverage is lousy, and also in Starbucks etc. I see from the RIM website it's shipping France too, so I guess there's an Orange Uniq variant as well. The press release mentions AT&T - but unless it's done (another) volte-face it doesn't have a currently announced plan to support UMA.
But the data support option is the more weird one - does this mean that you need to get your carrier's permission before you use it on you own WiFi network? So if I'm sitting in my office, connecting to my exchange server in my basement over my WiFi - or going out over my local ISP connection... I have to pay my operator for the privilege?! Maybe this is what the AT&T reference is about in the press release.
Also still unclear is whether the device is usable by enterprises for non-operator VoWLAN. Can I put a Cisco or Avaya softphone on it and connect directly to my IP-PBX via the Internet, as you can do with a normal Symbian or Windows smartphone? Is there an addressable 'naked' SIP stack? Unfortunately the full specs for the 8820 aren't on RIM's website yet.
It could be that it's only designed for operator-managed dual-mode IP-PBX based solutions using SIP, such as BT's Corporate Fusion offering. I could see AT&T playing in that space as well.
On the other hand.... it's started selling direct via Carphone Warehouse in the UK, which can activate the devices on 3 different operators' networks. Sounds like RIM may be edging slowly to carrier-neutral distribution after all.
Monday, July 16, 2007
I don't have time to go through all the ins & outs of this yet, but on first glance it looks like another step towards 3rd-party VoIP on mobile being given a chance to compete. Obviously there are arguments and counter-arguments on both sides, but from my point of view the fact that T-Mobile blocked access to numbers that their subscribers could reasonably expect to reach was a step too far. While I certainly appreciate the desire of operators not to subsidise WLAN-capable handsets & then see that subsidy used against them, the notion that they can just arbitrarily block outbound calls to number ranges they don't like is ridiculous.
And they made it worse by announcing to their users "You have dialled an incorrect number. Please check it & try again" - despite the inbound call coming from that very same number. If the recorded message had been honest and said "Sorry, you've tried calling a number from an operator we're still in negotiations/dispute with, please call from a non-T-Mobile phone" I would have cut them some slack.
On the other hand, if you want a well-argued contrarian viewpoint, Keith over at Telebusillis had this to say the other day
Friday, July 13, 2007
a) T-Mobile & Truphone are going to court on Monday about whether or not T-Mobile has to interconnect its normal voice customers to Truphone's shiny new UK mobile number range (delivered via VoIP rather than circuit-switched). I haven't waded through the respective legal documents in detail but T-Mo's lawyer wins the comedy prize for his comment on page 15 about the 'monopoly' of the newspaper seller.
b) Ofcom has ruled in a case about different interconnect prices for terminating on 2G vs 3G networks, relating to BT & the various other UK network operators. My reading of it is that BT has to pay retrospectively a 'reasonable' blended termination fee (higher than 2G) that incorporates the supposed higher costs of 3G voice termination. That was actually determined in Ofcom's document on interconnect earlier this year, but the dispute relates to an earlier period.
...but I'm trying to work out if BT's being really really clever on this..... despite its protestations about 2G vs 3G termination costs, it's actually in BT's longterm interest to have termination fees to mobile numbers being cost-variant across different technologies. This then gives it scope to argue the right to pay less for any future calls to mobile numbers that terminate on a cheaper network infrastructure - say over WiFi, or via a femtocell, via fixed VoIP, or to a voicemail server.
Monday, July 09, 2007
I can sort of see the argument - especially for 2G operators & MVNOs that don't have existing UMTS spectrum or built-out networks, and think that 802.16e might be cheaper or easier to deploy. Conversely, it may be that such devices would enable WiMAX startup operators to start off with lower capex if they concentrated on build-out in dense areas, and cut an MVNO deal for occasional rural use.
One problem might be indoor coverage, however - I can't see 2.5GHz or 3.5GHz WiMAX having great penetration in-building, which may mean WiFi is needed as well. Another thorny question is around authentication - although Comsys is positioning WiMAX as "a cellular-type technology" I haven't seen much enthusiasm for using SIMs with it thus far, although I guess that a UMA or VCC approach might work.
Nevertheless, an interesting option & worth keeping an eye on.
I'll take a punt that the underlying platform will support both - although possibly not in all operator-specific variants.
I suspect that T-Mobile US may have come up with a superficially compelling reason for a UMA variant for use at hotspots (& maybe employees' homes), although it's not likely to be much use when the device is inside a corporate WLAN & firewall. UMA isn't designed to work with complex switched WiFi networks (Cisco, Aruba, Trapeze et al). There have been various noises to make some of these UMA-friendly, but they don't appear to have got anywhere substantive. One of the problems is that UMA needs to make an outbound VPN connection to the operator's UNC - not something that the average enterprise security manager would be too happy about, unless a separate partitioned domain could be implemented on the wireless network.
Conversely, most of RIM's other operator customers aren't using UMA, and so presumably won't want to be spending money for lumps of complex software in the stack that will go unused. I suppose it's possible that they could ship with the UMA software 'dormant', though - especially if the per-device royalty was only payable on phones which were actually activated. Whether it'll have some form of VCC/pre-VCC client is another matter - I suspect we might have to wait for dual-mode BlackBerry V2.0 for that.
On the other side of the coin, I imagine that RIM has been keenly aware of the drive for SIP-based PBX-centric dual-mode devices - and conscious of Nokia's E-series and various Windows devices' incumbency there. The lure of unified comms, VoWLAN and possible tighter relationships with Cisco, Avaya, Divitas, Siemens, Varaha and assorted others must be pretty potent. I'd imagine RIM doesn't want to miss out on an emerging market for converged corporate voice/data devices.
But that opportunity comes with a sting its tail - although some of the fixed/mobile hybrid carriers are moving towards managed dual-mode IP-PBX implementations (eg BT Corporate Fusion), there is definitely a strong move towards enterprises buying 'vanilla' WiFi handsets through alternative channels, rather than operator-customised ones.
So.... could the dual-mode BlackBerry make an appearance through non-operator channels? Perhaps sold through a systems integrator, with the enterprise sourcing SIMs and voice/data airtime separately? It's got to be a tricky decision for RIM, given its strong carrier partnerships. But sooner or later, they have to deal with the undeniable fact that some enterprises will want to source all their own mobility hardware (including mobile email servers), and just pick one or more operators as a cellular bit-pipe.
Tuesday, July 03, 2007
Now, I should caveat my judgement in that I don't know the assumptions, definitions etc in the full report & model, but even so, this is so far beyond the "smell test" I don't know where to begin.
But lets start with:
- Enterprise - UMA initiatives are going nowhere
- Prepay (seen any FMC models suitable for 70% of the world's subscribers? I haven't)
- Inapplicability to CDMA/WiFi (ie about 20% of users)
- Limited base of broadband homes with WiFi
- Refarming 900MHz to give better indoor coverage
- Falling cellular prices making homezone pricing irrelevant
- SIP & VCC standardisation coming with 3GPP R6 and R7
- Lack of WiFi penetration into handsets
Some of this doesn't apply if the stats also include non-dual mode devices using UMA (eg femtos), but it still looks out by getting on for an order of magnitude. I haven't redone my full model & forecasts for a while, but if I had to take a punt, I'd be being generous if I said 10 million subs by 2012.
As always, let me know if you'd like a disruptive second opinion.
I'll try & update this post through the event, although I'm not sure if I'll be here tomorrow & Thursday.
One particularly interesting development has been the instigation of the Femto Forum and I've already heard a lot of discussion around standards.
But... I'm becoming increasingly skeptical of the near-term opportunity, especially in mature broadband markets. Some general areas of concern:
- rapidly-growing installed base of non-femto home gateways & settop boxes. In the real world, most early femtos will need to be "daisy-chained" off the back via an ethernet port, adding to installation and configuration headaches
- lots of unrealistic expectations about "single operator households" supposedly where one provider supposedly offers a whole family a bundle of mobile phones, broadband & maybe IPTV and so forth. Maybe in utopia, or in a country which is a mobile/broadband greenfield, but again in the real world, people have multiple phones, households have members with different requirements, longterm contracts are always out of phase, and so on. How do you manage a house with 2+ femtos?
- No mention thus far about how to optimise a femto offering for prepay subscribers
- By the time femtos are massmarket, 900MHz UMTS will have solved many of the macro coverage problems
- Ideally, handsets will be femto-optimised to enable them to use the indoor coverage "well", ie for data-related applications like content backup, TV watching etc
- Unclear what the working life of a femto is expected to be - and how that fits with the blistering pace of development of the rest of the IP/IT/WiFi/broadband part of the system
- Assumption that the femto voice-stream will get priority on the broadband connection. Who decides QoS prioritisation vs. home access to enterprise VPNs, streamed TV, home security or various other 'important' uses of the pipe?
- For mobile-only operators without their own broadband (or close ISP partnerships), the femto model assumes Net Neutrality, which has the potential to raise accusations of hypocrisy. ISPs will investigate ways to charge 'parasitic' mobile operators who want to use 'their pipes' for free.
None of these are unfixable problems. I still like the underlying concept of femtos. But these problems will take time & keep the new Forum occupied. I haven't done my own forecasts yet, but my gut feel tells me that one of my peer's much-ballyhooed 5-year forecast of 36m units is over-optimistic.