A few years ago, the first 3G UMTS networks to be launched involved some relatively simple choices.
In Japan in 2001, DoCoMo launched its FOMA service initially with just a single frequency band supported. This was 3GPP Band I, using 1920-1980MHz paired with 2110-2170MHz, commonly just called the 2100MHz or 2.1GHz band. Initial devices were launched without 2G dual-mode support.
In Europe, operators used the same 2.1GHz band, but were reluctant to launch devices which didn't also support GSM for fall-back. This typically meant adding support for 900MHz and 1800MHz 2G into the handsets, which was one of the contributory reasons for the delay in wide launch of 3G handsets, as well as their expense, clunkiness and poor battery size.
Nevertheless, at least the handset designers and operators knew what they were dealing with, when it came to frequency support - the 2.1GHz was a known quantity: challenging, but at least the goal-posts were fixed.
Now ratchet forward to the future introduction of LTE.
What's going to be the main launch frequency for LTE in Europe and Asia?
Is it the forthcoming 2.6GHz band, for which auctions are already starting? This has a fair amount of capacity, but may suffer from poor indoor penetration or the need for much denser networks of cell sites. There are also some issues around possible interference from WiMAX or other TDD technologies in the middle of the band, especially if certain regulators like Ofcom look to adopt national-specific bandplans with a unique mix of TDD and FDD.
Is it refarmed 900MHz GSM spectrum? It's a prime frequency, but not every operator has some of this to refarm. Will regulators insist on a reassignment as part of the refarm process? And will this be enough for LTE, especially as the broadband business case really needs 10MHz or 20MHz channels? There's certainly not enough for 3 or 4 operators per country to have 10, 20 or more MHz.
Could it be 1800MHz refarmed spectrum? Again a possibility, but not for everyone.
Maybe a straight replacement of existing 2.1GHz 3G?
What about the UHF Digital Dividend spectrum down at 450MHz? That's also patchy, and not available until analogue TV is switched off in different places at different times.
Then there's 700MHz spectrum in the US, and various other local variants in Japan. And what about China?
The bottom line is that nobody really knows. 2.6GHz is looking like a probable candidate, but certainly not the only one. And if handset manufacturers and chipset vendors have to start out with dual- or tri-band LTE, and also support 2G and 3G radios for fall-back, I think that the timelines and costs will suffer.