3G?s financial & technical booby traps
Britain?s recent spectrum auction will add a $35 billion up-front cost to the
delivery of 3G services. What?s more, it?s becoming clear that the technical
challenges of 3G will make network deployment a very expensive
proposition.
By Grahame Lynch & Tony Chan
hirty-five billion dollars for a technology that doesn?t exist yet. That?s what
five mobile telephony companies will be paying the UK government ? half now,
the remainder in 2006 ? for access to Britain?s third-generation (3G) 2-GHz
radio spectrum. The number is huge, some 10 times what the UK Government
had budgeted for. Many are wondering if Britain?s cellular companies have
taken on too huge a debt. The Times described the auction results as "the most
extraordinary poker game in British business history." The Financial Times was
blunter, describing the results as "folly."
Financial markets were also far from approving, with credit agencies Moody?s
and Standards & Poor?s both moving to rate Hutchison Whampoa, a successful
bidder, with a negative outlook. The British have recent memories of the CT-2
debacle, where entire national networks were trashed as they were overtaken
by GSM technology.
What?s more, the $35 billion spectrum bill is just the first cost of deploying 3G
networks. Still to come are the costs of new backbone infrastructure, new base
station sites, handset subsidies and employee retraining. The costs will at least
match those of 2G networks, and could end up costing more.
The auction madness doesn?t just end with Britain. Germany is about to embark
on its own spectrum auction, which some predict may result in $50 billion of
successful bids, given that Germany?s economy is a good third larger than Great
Britain?s. What might a US 3G spectrum auction raise in this circumstance?
$200 billion?
A W-CDMA network may require up
to 30% more base stations than a 2G
network.
The obvious concern is that cellular companies ? and along with them, their
financially codependent vendors ? will struggle with such a cost load. Some will
default, while others will struggle to get their infrastructure built out quickly.
Consumer prices for 3G services will be priced so high that their very viability as
a mass-market offering will be arrested. A new industry will be stillborn.
But such a scary scenario may not come to pass. Not all countries are going
down the highest-bid auction path. France, Japan and Hong Kong have all
elected to use a more conventional beauty contest approach to licensing. This
should hasten the viability of 3G services in those markets, creating
economies-of-scale for global manufacturing that can be leveraged in higher cost
countries.
Small chunks
What?s more, the $35 billion number is less formidable than it might look at first
glance. Spread between five licensees across the 20-year duration of the
concession, it will actually come to an annual outlay of $350 million per operator
? or $6 a year for each Briton. With well over a third of the British population
already using cellular phones, the economics may actually result in an equitable
outcome for both the public purse and private interests.
But $35 billion is a tremendous gamble for a technology that doesn?t really exist
yet. The five UK licensees will use the emerging wireless code division multiple
access (W-CDMA) standard, which combines a CDMA air interface with the
back-end platforms of global system for mobiles (GSM). W-CDMA exists
now, but in an extremely immobile form. Ericsson can demonstrate a system that
weighs 240 lb and is so big that it is housed in a minivan. The company
concedes that its test system is so power hungry that it would chew up a
standard cellular battery in about one second.
Leading proponents of the technology say that big advances are imminent.
Japan?s NTT DoCoMo still predicts that it will rollout a W-CDMA system next
year, albeit using a local Japanese FDD interface rather than the TDD interface
specified internationally. It has been testing W-CDMA since 1998. Motorola
says that the first W-CDMA system to incorporate both a physical and signaling
layer could be ready as soon as this month.
Interestingly, the immediate driver for W-CDMA, and the cdmaOne rival
standard ? CDMA 2000 ? is not so much wireless Internet and video, but the
desire for increased voice capacity.
NTT DoCoMo supports well over 30 million customers on its time division
multiple access (TDMA)-interfaced network. With the success of its 9.6 kbps
i-Mode service, it is running out of capacity and has even halted new
subscriptions in some locations.
W-CDMA offers an immediate attraction ? the ability to support 207 voice
channels per sector, versus 100 for a typical TDMA or GSM network. NTT
DoCoMo will also be building its W-CDMA network separately from its
personal digital cellular (PDC) network, ensuring that there are no tricky
accommodations of both 2G and 3G users on the one set of infrastructure.
Similar capacity improvements are in store for cdmaOne networks. Ericsson
says that a CDMA 1X upgrade should double voice capacity in each
1.25-MHz carrier. Use of wider spectrum bands will expand this capacity
further.
More base stations
But in W-CDMA?s greater capacity lies its trap ? its maximum data throughput
of 2 Mbps per cell is highly theoretical and unlikely to be replicated in a
real-world environment.
W-CDMA operates between 1.8 and 2.1 GHz. In theory, W-CDMA should
be an easy fit for existing cellular networks in the 1.7- to 1.9-GHz band. 3G
cells could be deployed at existing 2G sites with little trouble. But Nokia 3G
Marketing Director Timo Poikolainen says that Nokia tests show that a
W-CDMA network based on a GSM-1800 deployment can only support
maximum cell hand-off rates of 144 kbps. Faster performance will require more
base stations.
A 3G packetized network should cut
backhaul and trunking costs.
A W-CDMA network that seeks to leverage off a 900-MHz network will fare
worse, since networks in lower bands typically use fewer base stations. The
trade-off will be either lower capacity or the need for more cell sites.
The coverage range of a W-CDMA cell theoretically compares well with that of
a TDMA or GSM 1800 cell. But using a typical 3G spectrum allocation of 2 x
15 MHz, a W-CDMA network may require up to 30% more base stations than
a 2G network.
Lower data rates
Another problem affecting data rates is the limitations of handset technology.
According to Nokia, handset power limits restrict uplink data speeds to 144
kbps for all but the smallest distances. To achieve rates of 384 kbps, a
W-CDMA network would require 80% more base station sites than a 2G
network.
What?s more, the bandwidth capabilities of a W-CDMA cell are shared among
all users. In an optimal situation, a user should be able to achieve 384 kbps in a
mobile environment or 2 Mbps in a localized environment. The use of a 5-MHz
carrier on a 15-MHz allocation means that each user should, ideally, be able to
access three cell sites. The downside? If there are more than three simultaneous
users, the data rates fall from their optimal levels.
Such sharing allows for greater network efficiencies than the circuit-switched
alternative. Says Motorola?s Ching Chuang, "In today?s 2G network, once a
channel is taken, it is occupied and no one else can get on it. But with 3G,
bandwidth will be shared, so the network will be able to support many more
users at the same time."
However, this doesn?t necessarily enhance the user experience. If initial user
experiences with 3G data speeds are underwhelming, it will be difficult for
operators to price services at a premium to existing offerings. Ironically,
circuit-switched data offerings may become premium services if W-CDMA
really takes off ? simply because they can provide guaranteed quality.
3G?s savings
But there?s one aspect to a 3G network that should have operators cheering.
The move to a completely packetized network should cut backhaul and trunking
costs, which currently account for up to 25% of the total cost of a cellular
network. Several vendors have made an attempt to quantify these savings.
Nortel suggests that the cost of transferring a megabit over a network could fall
from 20 cents currently, to 11 cents in 2002 and 4 cents in 2004 through using a
packetized platform.
Operators may also find that they can leverage off existing 2G networks through
2G/3G dual-mode handsets. 3G W-CDMA nets can be deployed in high-usage
areas, with customers roaming onto 2G networks elsewhere that support
technologies such as general packet radio service (GPRS) or cellular digital
packet data (CDPD).
But these may be small mercies in a difficult market. 2G operators, especially
those with no experience of CDMA deployments, may find themselves
struggling to make the technical transition to 3G.
It?s worth remembering that 2G CDMA operators have experienced their own
fair share of difficulties implementing a complex technology. Sprint PCS CEO
Andrew Sukawaty says that the CDMA standard had "many teething issues" as
recently as three years ago, while in 1998, Hong Kong?s only cdmaOne
network came dead last in a independent network quality study against some 10
other GSM and TDMA networks.
Not only will 3G operators be overlaying a radically different air interface in
networks optimized for a TDMA system, but they will also have to deal with the
unknown quantity of shifting usage patterns. They will need more base station
sites. They will need to subsidize 3G-enabled handsets, which may have an
early real cost of as much as $700 to $800 per unit. They will also need to
manage consumer expectations, which have been built up so high that they will
inevitably be let down.
The biggest problem may be one out of the individual operators? control. If one
of the early bidders for 3G defaults on a payment, then the reaction of the
finance community will condemn the entire 3G push. Too many people have big
memories of another recent billion-dollar mobile industry debacle ? Iridium. |
