Speaking Engagements & Private Workshops - Get Dean Bubley to present or chair your event

Need an experienced, provocative & influential telecoms keynote speaker, moderator/chair or workshop facilitator?
To discuss Dean Bubley's appearance at a specific event, contact information AT disruptive-analysis DOT com

Wednesday, April 09, 2008

Is WiMAX losing the battle to 3G for connecting laptops?

I've been looking at the 3G vs WiMAX battle from a number of angles recently, for some work I'm doing about spectrum. And I've also been layering on some analysis about how this fits with the continued explosion in mobile broadband for PCs, especially with HSPA dongles.

The battle seems to boil down to a few sets of variables:

  • Can Intel get integrated, WiMAX-enabled laptops into the market as fast as it did for WiFi?
  • Can mobile operators encourage (subsidise?) laptop manufacturers to embed 3G modules instead of (or as well as) WiMAX?
  • Is the current laptop installed base going to churn as fast as did (about 3 years) when Centrino was introduced?
  • While existing HSPA dongles & inexpensive mobile broadband subscriptions are flying off the shelves, can the networks scale fast enough (in terms of backhaul & maybe extra sites) to avoid worsening user experience?
  • What happens when a laptop has both WiMAX and 3G in it/attached to it? How does the connection manager cope and assert preferences?
  • Are 3G operators' existing frequency allocations (up until now underused) soon going to be struggling to deliver a decent broadband experience to a massmarket of PC users? What frequencies are available to them, and in paired (FDD) or unpaired (TDD) spectrum?
  • How well can different operators deal with supporting pre-paid models for mobile broadband rather than monthly subscriptions?

Clearly, the 3G guys have been a bit rattled by WiMAX's successes in spectrum allocations. This doesn't relate to specific deployments (or delays) like SprintNextel's, but by the strategic success of the lobbying machine when it comes to important decisions like the ITU's acceptance of WiMAX as part of the IMT family of allowable technologies in 3G-designated spectrum. Add to this the recent movement in Europe towards supporting "technology neutrality", especially in 2.6GHz, and the recent endorsement of a report (CEPT #19) permitting national regulators to deviate from the supposedly-harmonised original band plan for 2.6GHz to permit more WiMAX-suitable TDD spectrum to be allocated, at the expense of 3G/LTE-optimised FDD.

Ofcom's new statement on the upcoming 2.6GHz auctions clearly demonstrates that it expects to push ahead with a flexible band-plan approach, and although some other EU regulators (Sweden & Austria) have toed the conservative 3GPP-friendly CEPT 05(05) line, I suspect others will follow Ofcom's path.

Conversely, it looks to me as if the WiMAX community is getting rattled by the seeming success of mobile broadband offers by the HSPA and EVDO operators. The whole 'dongle' phenomenon has sprung up in a narrow window of opportunity before WiMAX-enabled offers start to emerge widely. They're also supported by existing MNOs' huge retail and marketing presence, and arguably are setting price points that are making future new entrants wince as much as incumbents. And there are some early and very smart moves to support prepay for data.

I had a very persuasive briefing call yesterday with a leading WiMAX advocate who felt that existing 3G networks were going to saturate very quickly with the current consumer dongle offers, and felt that without extra capacity the propositions only had a limited shelf-life & appeal. While I think that some extra capacity may come from femtocells I don't see it changing the dynamics fast enough to make a huge difference - I reckon some of the European 3G networks will start to creak by the end of this year. As a side note and possible anecdotal evidence, my HSDPA modem on 3 UK now seems to struggle to get much above 300kbit/s, and it's often much worse.

So where are we? How will laptops be connecting in 3-5 years' time?

The future situation for laptop-suitable WiMAX spectrum is definitely improving, in some places more so than others.

But 3G operators have stolen an early lead in getting devices to market, although there are storm clouds gathering about scalability of their existing networks.

And as I've written about before, 3G-embedded PCs are still going nowhere, although ultra-cheap laptops (subsidisable down to £zero) may make a difference for those operators who have sufficient retail skills to sell & service computers.

And I certainly wouldn't write off WiFi, as it becomes more ubiquitous, easier to use, and increasingly free.

I think that the pivot point will be the perceptions of quality of existing 3G dongle-based service over the next 12 months as the networks fill up - and whether the existing mobile operators are getting sufficiently good margins to make them want to bid big for 2.6GHz spectrum, even in the face of a likely Intel-led onslaught of WiMAX-capable PCs.


Anonymous said...

But Dean, let's put the horse before the (Intel-sponsored) pile of horses*it in the cart ;-)

It's pointless putting a WiMax chipset into a laptop *if there are no basestations nearby to use*.

WiMax is no more efficient (in bits/Hz/sec) than HSPA. Everyone agrees. Even the WiMax Forum (somewhat begrudgingly).

Even if you get the spectrum & hardware for free, the other 9/10 of wireless network CAPEX is still required. Sure, put up a handful of WiMax BTS in London on top of ISP POP's on the cheap, but even at 450MHz (assuming digital dividend goes that way) your link path is still constrained by device output power. And with say 20MHz to play with (even 100MHz!) you will hit the wall after about a few hundred subscribers.

No, if you want to play the mobile internet game, a seat at the table will cost you a few £Bn in network rollout/backhaul/rent/O&M/CAC.

Apart from xMax (remeber them ;-) there is still no free lunch on the horizon.



Brough said...

Unlike WiFi, WiMAX follows the same business path that 3G follows, i.e. service providers using cell sites that support many users. Even if WiMAX is technically ahead of 3GSM (an argument that's secondary), WiMAX has the same cost for spectrum, similar costs for infrastructure and the same problems (and costs) for backhaul.

The biggest difference is volume. There are more than a billion GSM chipsets made per year (and the chip manufacturers can count on similar 3G chip volumes in due course). There are over 300 million WiFi chipsets made per year and this is growing rapidly.

Unfortunately, there are less than 1 million WiMax chipsets manufactured per year and no matter what Intel does, this isn't going to change materially any time soon.

So in the end, the best WiMAX can hope for is that it can achieve & maintain technical leadership and use that leadership to match what Qualcomm did for many years with CDMA.

Evert said...

While your post is very valid there is always the enigma of stupid management: http://wimaxxed.blogspot.com/2008/04/o2-uk-is-obviously-run-by-bunch-of.html

Ram said...

I agree that it will become increasingly difficult for WIMAX operators to compete with flat-rate USB dongle offers as low as 30 Euros per month. The 3G operators would have written down significant capex costs by the time WIMAX operators start deploying BTS. Can WIMAX operators survive as a pure mobile broadband operator without SMS and voice cash cows? Seems unlikely!

Dean Bubley said...

Thanks for the comments.

Mike - was it pointless putting a WiFi chipset into laptops, when there were still very few access points nearby to use? If you're in Intel's position, you're able to just buy some chickens, rather than waiting for the chicken/egg problem to solve itself.

The point I was making about spectrum was that it means a greater number of Mobile WiMAX providers will be able to start deploying networks over the next 18 months. Sure, it'll take time to roll them out, but if I was a WiMAX equipment vendor I'd be pleased that there's now scope for perhaps 2-3 national networks per country rather than just 1.

Totally agree though about the capacity issue - which is why UK operators with 2x15MHz of 3G spectrum are already struggling to keep up with demand, and you have unbelievable stories like O2's EDGE network being faster than HSDPA.

I'm not so convinced that the entry price is as high for WiMAX operators as for 3G/LTE. (I haven't heard many people suggest the emergence of lots of greenfield 3G operators in upcoming auctions).

Usage patterns for PCs are different to phones, so I suspect that coverage can scale more slowly (helped by probable rollout conditions on 2.6GHz spectrum licences too). Also, PC users tend to be happier with "input your postal code to see if you can get service".

Also, WiMAX BTS aren't stuck with the need for legacy TDM or ATM access connections. I also suspect PC usage hotspots coincide quite nicely with places which have fibre available for backhaul too.

On the other hand, 2.6GHz indoor penetration is definitely an issue.


There is a real risk of comparing apples & oranges here. In particular, there are broad differences between phone-centric value chains (eg handset chipsets & manufacture & distribution) vs non-phone wireless devices like dongles & laptops themselves.

In particular, laptops are not (generally) sold through carrier channels, or sold in carrier-specific variants and with a few minor niche exceptions I don't see this changing much.

Laptops are also used in very different ways to phones. In particular, they don't need unique & permanent identifiers (phone numbers). And apart from the specific exception of onboard trains, full mobility is much less important.

So fripperies like SIM cards & much of the expensive mobility IPR is wasted on 3G laptops. I'll also bet that WiMAX operators will have much lighter or more centralised testing & certification processes for PC manufacturers, and not require each device to spend months in each operator's labs.

In addition, PCs (whether used with embedded chipsets or dongles)

Ram - 30 Euros? Now down to £10 / €13 in the UK - and I even saw one store with a special offer at half that price. The question is whether those prices are sustainable for a decent level of customer experience as the networks load up. I'm not convinced.

Overall, I should point out that I'm not a particular WiMAX fanboy. It certainly has its problems, as it's coming from a standing start. But 3G also has some issues, and LTE isn't here for quite a while yet. Living in the UK, I've got a competitive range of both fixed & mobile broadband, so I'm not as desparate for extra choice as someone in the US. So my whole point is that the picture isn't clear - at the moment, in much of the world, HSDPA is the answer for PC-based mobile broadband. But that might start to swing back on an 12-24 month view.

Anonymous said...

Dear Dean,

Don't assume that the capacity of a WiMAX network will be any higher than that a 3G network will have at the time a WiMAX network will be operational, the reality today is that the only operational mobile WiMAX network, Wibro in Korea has equal or even lower spectral efficiency than today's HSPA networks and does not offer higher bitrates.

Don't assume that the 3G networks will fill up any time soon, the reality is that all 3G HSPA networks are using only a fraction of the available capacity, i.e only one carrier out of typically 3 is used and at only a fraction of all sites you can see a need to upgrade to a 2nd carrier any time soon. After using all available carriers some time in the future there are several other upgrade paths to follow and it is therefore wrong to assume that there will be a capacity limitation in the foreseable future.

Don't assume that your disappointing experience of HSPA depends on a system capacity limitation as the limitation in your case most certainly is the transmission capacity to the base station site, most radio sites in Europe use E1 ATM links which provide 2Mbps to be shared by all users. It is of course needed to upgrade the transmission to each site if you want to provide a better user experience. All 3G network vendors today offer transmission solutions based on either wire-based IP or microwave solutions. My own experience is that I get 5.7Mbps downlink and 1.25 Mbps uplink and that is down in my cellar! but my provider has aleady upgraded the transmission to most sites. It goes without saying that this transmission issue is as valid for a WiMAX network (something that Sprint recently reported about).

So please Dean, don't assume!

Anonymous said...

Thanks for the detailed reply Dean & others. interesting.

Dean, you point out all the things that WiMax *doesn't* need, that an LTE BTS will. I question this logic....

you are asking the city to fund a business model where:

It's only where you think people want to use laptops (nomadic vs mobile)
You are not offering voice or text - basically laptops ONLY then.
...and you'll have to sit outside to use them. Unless you offer in-building coverage. Order of magnatude harder/costlier.

I don't see that having the chipsets in some laptops would make me happy to lend a few hundred million to build even a smattering of sites. I managed a rollout of TDD UMTS, where there was board pressure to get the coverage map and therefore supposed addressable market as big as possible as fast as possible. wheel in RF lead to say "need more sites boss" and the response is 'fixed budget, make do'. Result? Crap indoor coverage, pissed off customers, 50% return rate.

Customers now expect that their mobile broadband is just that - mobile. They will NOT pay for one provider to be mobile and another to be nomadic. Agreed for laptops, mobile doesn't need to have handoff to have value, just be ubiquitous. Nomadic? Remember Rabbit?

And as a few others point out, by the time you've planned & built your even small network, HSPA will be into Rel 7 with speeds of 42MBps.

The 'build it (in) and they will come' model doesn't apply. The wiMax emperor still has no clothes.