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

Friday, November 13, 2009

Operators should push vendors & silicon companies to support 2.6GHz HSPA+ as an alternative to LTE

Following up on my post from two weeks ago about HSPA / HSPA+ in 2.6GHz, I'm now more sure that this is a becoming a wise option - and operators and device vendors should start pushing their suppliers to support it. I'm becoming increasingly sceptical about the short-term case for LTE, especially for GSM/UMTS operators in developed markets like Western Europe.

The received wisdom suggests that "new spectrum = new technology". I can certainly see the appeal and elegance for radio network engineers to put the newest, shiniest kit into the spanking-new bits of frequency in 2.6GHz and the digital dividend.

But I think that for current HSPA operators, they should think twice. Even the theoretical gains in efficiency for LTE vs. Release 8 or Release 9 HSPA+ are relatively modest. With 64QAM and 2x2 MIMO, I've heard figures of 20-30% *might* be achievable. However, given views of contrarians like Moray Rumney from Agilent, these gains may well not be attainable in the real world - or at least, only for a certain proportion of the time under specific user/cell scenarios. In terms of actual *average* throughput in normal usage, there may well be only a wafer-thin margin between them.

Yes, LTE has better latency (in theory), a flatter and maybe cheaper network, and the ability to use thin slivers of spectrum.

But this needs to be set against the need to run LTE as an overlay on HSPA anyway (3 sets of network opex....), plus the extra cost and complexity in handsets, the huge testing and optimisation costs, the probable flaky hand-offs between LTE and 3G/2G, the ongoing issue of voice support and numerous other unknowns. Add to this the fact that LTE does not appear to offer any obvious new business models compared with HSPA+ (especially if only used in limited "hotspots") and the business case dissipates even further.

That's not to say LTE won't be improved - after all, HSDPA has proven a superb "bug fix" for WCDMA, only 5 years after it was introduced. Prior to that point, 3G was pretty pointless - in hindsight, operators would have been better off leaving the 2.1GHz spectrum unused, or perhaps temporarily putting EDGE in it, although under old regulatory regimes that probably wouldn't have been allowed.

This time, there are more options.

I think that operators should give serious consideration to the scenario of continued upgrades to HSPA, including putting it in new spectrum bands like 2.6GHz. Much of the new radio equipment could be easily upgraded to LTE at a later date - if required. There's possibly an argument to skip current LTE entirely and wait for real 4G - LTE Advanced - as an eventual migration path.

Given that operators are currently starting to put HSPA in refarmed 900MHz (surely also a "new band" effectively?), why not also 2.6GHz?

Either way - I think that operators should start leaning on their suppliers, both infrastructure- and device-side - to support HSPA2600 as an option at least.

Edit: any game-theorists reading this might also want to ponder on the competitive impact of those most benefiting from early scale economies for LTE devices (on any band). In particular, CDMA operators moving to LTE will likely be disproportionately affected if their HSPA peers don't follow suit at about same time. Will LTE enable a marketing win vs. HSPA+? Given that neither is 4G I can't really see why - 3.75G vs. 3.9G isn't really a great headline-maker. In fact, if WiMAX and LTE operators can use 4G as a branding device, I see no reason why HSPA+ operators shouldn't too.


Arun Demeure said...

Realistically I'm a bit skeptical this is going to happen because operators still seem quite optimistic about LTE. If something bad does happen, I doubt they'd realize it in time to influence anybody. Maybe I'm a bit too cynical but heh...

What I find more interesting is the possibility that this might give a boost to software defined radio RF companies, such as Bitwave. They haven't had a lot of success so far but they (Bitwave and Lime Microsystems for example) have real products out today and this might give them a small opportunity to break into the market.

Another thing is that several companies are moving their RF design to be much more digital-centric. Two great examples of that are Infineon and Icera (the latter obviously also applying SDR for the baseband with many many real design wins) - it is possible that their next-gen designs (both will be on 65nm) are more flexible than they are willing to reveal right now, and might actually be capable of 2.6GHz without any change. Or maybe not, but it's worth pondering. Others such as Qualcomm are definitely not as aggressive on that front FWIW.

This doesn't solve the infrastructure side of course, which I am not as familiar with.


Moray Rumney said...


The potential for SDR to make a difference in the baseband continues to grow. However, the problems in developing multi-band multi-format handsets goes way beyond the baseband. I have yet to meet a digital PA or a digital antenna (although research continues), and dealing with these two characters is a major headache for device designers. There are only so many things that can be economically intgrated into a single device. On top of the device design challenge there is the operational challenge of multi-band multi-format devices. As the number of potentially simultaneous interfaces grows linearly, the number of interoperability scenarios grows exponentailly, and every one needs to be designed, tested and optimized if the holy grail of seamless connectivity is to be maintained.

The days when it was obvious what to develop have long passed. GSM entered the market with a signle band and two services - voice and SMS. UMTS/W-CDMA followed 10 years later with a similar footprint. LTE on the other hand has an entry point that covers perhaps 20 FDD frequency bands, 8 TDD frequency bands and interworking scenarios for GSM, W-CDMA, HSPA, CDMA2000, 1xEV-DO, WiMAX... This fragmentation can mean only one thing in the short term and that is high cost, high complexity and probably questionable performance. Maybe in a few years the viable options will reduce allowing the industry to focus on and optimize a smaller subset of technology but nothing I am seeing yet in the industry is actively driving towards that scenario - for the time being it is all about what is possible rather than what is commercially viable.

In such an environment it will be economics that determine the winners and losers. Some amount of distortion of the market will exist for a while for those with deep pockets but ultimately, the market will merge towards good enough low cost solutions.


merlen hogg said...

Hi Dean,
I loved reading this piece! Well written!

Merlen Hogg