Too little, too late
James Tulloch,
Mobile Communications International
Issue 95, 01 October 2002
'Build the network, and they will come', is the 3G dream. But, despite recent UMTS launches, the signs are that there is very little network out there.
When is a launch not a launch? When it is a UMTS launch, judging by Sonera's 3G-type-services-over-GPRS fudge, Mobilkom Austria's network-now-services-later approach, and Hutchison 3GUK's rollout to 1,000 'friendly users'. For genuine commercial launches we will, in all likelihood, have to wait until next year.
Tradition dictates that the 'lack of handsets' is to blame for 3G delays. Both Sonera and Mobilkom raised the issue. But perhaps lack of infrastructure is equally culpable. Establishing exactly how many UMTS base stations an operator needs falls into the how-long-is-a-piece-of-string category.
It depends on multiple trade-offs between coverage, capacity, throughput, and QoS. When Chris Gent forewarned that Vodafone's initial UMTS throughput would be in the 64kbps range that gave an indication of what has lately been confirmed by infrastructure vendors: very little UMTS kit has actually been deployed, and revenues booked from 3G deals are negligible.
Nokia, which claims 35-plus WCDMA infrastructure contracts worldwide, had by the end of Q2 deployed €686 million worth of 3G infrastructure, according to Rene Svendsen-Tune, VP, marketing and sales. "Until we have technical proof, we don't recognise revenues of new equipment," he said.
Ericsson presumably has similar issues, as it will make a loss on 3G this year. J-Phone, for example, postponed its 3G launch from June to December. Officially, this was due to unstable WCDMA standards. But a month earlier, Ericsson was reportedly straining to meet J-Phone's delivery deadlines.
Siemens, meanwhile, has deployed 'several hundred' 3G base stations for its 18 customers in Europe and Asia, according to Joachim Horn, senior VP, sales. Thousands more have been delivered and will be in place by the end of the year, he claims.
These seem to be remarkably small numbers. European 3G operators were expected to spend the cost of their licences (€115 billion) again on infrastructure, full coverage requiring thousands more BTSs than GSM. Yet Mobilkom has reportedly spent only €72 million to launch its network to 25 per cent of the Austrian population, suggesting a serious coverage-capacity-throughput trade-off. In a filing made by the German government to the ITU, the network requirements for a single operator were simulated for three 60km x 60km areas: Berlin (metropolitan), Frankfurt (industrialised region), and Münster/Osnabrück (remote/rural). The study estimated that the total number of cell sites required for the three areas would be 15,944.
Which brings us to our piece of string. Of those cell sites the vast majority are micro and pico cells, which only appear later in the deployment cycle. In the classic model of cellular deployment, operators first erect a few tall antennas, put lots of power through macro cells, and see what happens. BT Cellnet started GSM operations with 500 BTSs. Micro cells come later if and when the networks start to get crowded, and pico cells when people start complaining about indoor coverage and so on. There is no reason why UMTS should be any different.
Realistically, nobody is likely to achieve national UMTS coverage, or even want to. Apart from Sweden, UMTS licences in Europe stipulate not geographic coverage but coverage based on population. In any case, operators may have planned their networks, and will continue to acquire base station sites, the least expensive part of roll-out, but they are delaying investment.
Deployments are being driven, or held back, not by regulatory obligations, but by economics. Orange Sweden, for example, recently complained that obstructionist local legislators would make it impossible to fulfil 3G roll-out requirements.
But blaming it on restless natives is unsatisfactory. More revealing, perhaps, are the concerns that Orange lodged with the Swedish National Post and Telecoms Agency. The crux of these, suggests a research note from Nomura, was that 3G base station amplifiers are so inefficient that it would be uneconomical to roll out a network to support a large number of subscribers. Nokia says it will have more efficient amplifiers in 2004.
"Therefore, if you are an operator, there is little need to roll out 3G until then," concludes Nomura. Orange asked the Swedish regulator for a reprieve until 2006, but this request was denied.
The uncertainty in Europe has done nothing for WCDMA's prospects further afield. In Australia, Telstra Mobile's managing director David Thodey recently floated the possibility of adopting CDMA2000 1X. Even more worryingly for the UMTS lobby, there are signs that China Mobile is wavering on its commitment to WCDMA.
For all these reasons, the EC is extremely keen on network sharing. Last month it gave its blessing to plans by O2 and T-Mobile to share capacity in the UK. Anything to get 3G off the ground. And would all the major vendors favour operator consolidation, despite the apparent damage it would do to their order books? "We would welcome it," says Siemen's Horn, and Svendsen-Tune echoes him, while Ericsson has publicly stated that the market is 'overcrowded.' A few robust contracts are better than a lot of sickly ones.
telecoms.com
==========
"Nokia says it will have more efficient amplifiers in 2004."
Like NOK's wCDMA dual mode was "real" and working in March?
Like NOK launched 3G with Sonera in September?
Like NOK promised US EDGE carriers handsets in 2002?
How much infrastructure - what cell density - is really required for viable GSMGPRSEDGEEGPRSUMTSwCDMAHSDPA evolution?
ROTFLMAO |
