The Great IP Debate

[David Farber <dave () farber net>]

Begin forwarded message:

From: Dewayne Hendricks <dewayne () warpspeed com>
Date: July 11, 2004 7:56:36 PM EDT
To: Dewayne-Net Technology List <dewayne-net () warpspeed com>
Subject: [Dewayne-Net] The Great IP Debate
Reply-To: dewayne () warpspeed com

The Great IP Debate

While the conceptual merits of IP networks are impressive, the economic  
viability and technical feasibility for wireless networks remain  
questionable.
By Sue Marek
July 1, 2004
Wireless Week
<http://www.wirelessweek.com/article/CA432357? 
spacedesc=Features&stt=000>

Telecom visionaries have long painted an idyllic picture of the  
converged wireline/wireless services that will one day be possible as  
operators migrate to all IP-based networks. The allure of all-IP  
networks is compelling because at the IP layer, both voice and data  
traffic are transported as packets of data, allowing carriers to  
efficiently deploy new services such as real-time multimedia with  
integrated voice and video. And once both wireline and wireless  
networks get to a common IP platform, telecom carriers could seamlessly  
merge their services and offer customers a truly bundled package.

But at the same time, wireless operators are somewhat leery of the  
IP-based services that these 3G networks can accommodate, particularly  
VoIP. The technology offers some advantages when it comes to deploying  
advanced services or managing the network elements; having an all-IP  
network instead of separate voice and data networks means that carriers  
have fewer pieces of equipment to deploy and maintain. Yet industry  
insiders say that VoIP, in some instances, is a less efficient  
technology than traditional circuit-switched voice networks, which have  
been optimized to use spectrum very proficiently.

VoIP or Circuit Switched? "All-IP networks in theory sound great," says  
Jane Zweig, CEO of The Shosteck Group. "But with the particulars of  
radio on the mobile side, migrating to VoIP requires over-provisioning  
of the network."

For example, in an IP environment, voice packets contain a lot of extra  
information, called headers, to ensure they are routed to the right  
location. These headers take up a lot of bandwidth and therefore are  
less efficient — and slower. To maintain service quality, voice packets  
must be delivered in a timely manner or else they will suffer some  
degradation.

In addition, carriers have invested so much in their legacy voice  
networks that many hesitate to move voice traffic from the legacy  
network to an IP platform. "Circuit switched is not going away. It's a  
legacy system that will be with us for a long time," says John Marinho,  
vice president of marketing and offer management at Lucent's Mobility  
Group. "But VoIP is the cornerstone that will enable a lot of things to  
be possible in the next several decades."

Exactly what voice services will be used for VoIP and what will remain  
on the circuit-switched network is uncertain. However, carriers are  
putting VoIP on their migration paths. "VoIP is part of the long-term  
solution" for wireless networks, says Rod Nelson, executive vice  
president and chief technology officer of AT&T Wireless. "It offers  
lots of potential for advanced services such as PTT or multimedia calls  
with video and voice. But for most normal cellular voice traffic, it  
doesn't offer huge economic gains compared to what has already been  
built into the existing system."

The Path to VoIP From a technical standpoint, operators could deploy  
VoIP in current-generation networks to provide high-quality service.  
But observers say wireless networks need a lot of bandwidth and a  
bigger uplink than existing UMTS Release 99 technology or the current  
iteration of 1XEV-DO, Rev. 0. If carriers migrate to the next versions  
of these technologies — high-speed data packet access (HSDPA) or  
1XEV-DO Rev. A — uplink speeds will increase, making VoIP much more  
viable.

On the GSM side, Cingular Wireless recently issued a request for  
proposal to the vendor community for UMTS/HSDPA infrastructure, which  
it plans to deploy across its network. The carrier is testing UMTS in  
its home market of Atlanta using Lucent gear; the HSDPA component  
involves a software upgrade to the UMTS equipment to support data  
speeds of up to 14.4 Mbps downstream. According to Lucent's Marinho,  
HSDPA provides speeds of 384 kbps upstream, which is still limited but  
will support VoIP. Marinho adds that another UMTS uplink solution in  
development, called enhanced uplink data channel (EUDCH), could give  
GSM operators 4 Mbps in both directions, making it very compatible with  
VoIP.

Still, Cingular's deployment of UMTS and HSDPA technologies doesn't  
necessarily mean it will automatically migrate to VoIP, according to  
Kris Rinne, acting chief technical officer of Cingular Wireless.  
However, Rinne says that IP-based services are appealing. "IP services  
are much more efficient regardless of whether it is in radio spectrum  
or backhaul transport," Rinne says. "It's more efficient than circuit  
switched, but you must constantly balance that efficiency with  
quality."

On the CDMA side, 1XEV-DO Rev. A has an uplink speed of 1.8 Mbps with  
3.1 Mbps downstream, making it possible to send voice packets with  
header information over Rev. A and not cause degradation of voice  
quality. Although no U.S. CDMA carrier has announced that it was  
testing or planning to migrate to Rev. A, Verizon Wireless Chief  
Technical Officer Dick Lynch said at CTIA Wireless 2004 that it was  
possible the company would continue on its EV-DO path and perhaps at  
some point offer a type of voice service over EV-DO.

Joining Verizon on the EV-DO path is Sprint PCS, which earlier was  
examining EV-DV as its preferred choice but late last month formally  
announced that equipment availability and market timing prompted it to  
forge ahead with 1XEV-DO. Sprint could decide to implement 1XEV-DV as  
it is building out its 1XEV-DO network, but DV equipment is not  
expected to be available until 2006. Although 1XEV-DV offers both a  
voice and data component, the voice portion of EV-DV is circuit  
switched. According to Peter Jarich, senior analyst, wireless  
infrastructure at Current Analysis, EV-DO Rev. A gives DO more  
potential as a voice network. "It's too early to say, but DO does seem  
more compelling," Jarich says.

Timetable For VoIP Though the debate over VoIP vs. circuit switched may  
be heating up among the carriers and vendors, VoIP over wireless likely  
will not be a reality for some time. According to Kai Konola, director  
of strategy and business development at Nokia Networks, one form or  
another of VoIP will migrate to cellular networks in the next four or  
five years. However, the mass market won't see VoIP over cellular for  
some time. "The capabilities of the radio networks to support VoIP in  
mass scale is not optimal," Konola says. "It makes more sense for  
carriers to provide voice over circuit switched."

The migration from circuit switched to VoIP likely will be gradual.  
According to Konola, many other factors are involved such as having  
terminals capable of handling VoIP, which require a lot of processing  
power.

Still, once VoIP does make it to the cellular network, most vendors  
believe that at some point the circuit-switched network will go away.  
"People will outgrow the legacy network," says Bruce Gustafson,  
director of strategic marketing, wireless networks at Nortel Networks.  
And as carriers invest more in the packet network, gradually the mix of  
subscribers on the IP core vs. the circuit-switched network will  
change. However, Gustafson also sees that different carriers will make  
the move to VoIP at different times based on their business strategies.  
"It also depends upon how wireless standards are implemented,"  
Gustafson says.

Others are less bullish on VoIP being offered via the cellular network,  
primarily because of the business case. "The benefit of doing VoIP over  
the air interface isn't that clear," says Lars Nilsson, director of  
business strategies at Ericsson. "The systems are not there."

Others echo that sentiment. "The economics of VoIP in wireless are  
different than wireline," Nelson says. "I don't know that the same  
drive is there, particularly for implementers of UMTS, which brings  
into it integrated voice capability. If the technology you are  
deploying doesn't have a voice solution, you would be more motivated to  
do VoIP. With UMTS, it's less urgent."

Gustafson says carriers already are implementing VoIP in the core of  
their networks to backhaul traffic. Exactly where the carrier converts  
the traffic from circuit switched to VoIP depends on the network  
topology. The benefit, according to Gustafson, is carriers can  
transport more traffic using VoIP in the core. "Instead of having a  
half-used switch here and there, we are seeing carriers ask for one big  
switch that will carry 50 percent voice traffic and 50 percent data  
traffic," Gustafson says.

VoIP may be garnering industry buzz right now, but how it will actually  
make its way to the cellular network is subject to debate. For wireless  
operators to realize the vision of truly converged services, they need  
to move to an all-IP based network. But significant questions remain as  
to whether it will be economically and technically viable for the  
long-term.

Archives at: <http://Wireless.Com/Dewayne-Net>
Weblog at: <http://weblog.warpspeed.com>

-------------------------------------
You are subscribed as interesting-people () lists elistx com
To manage your subscription, go to
 http://v2.listbox.com/member/?listname=ip

Archives at: http://www.interesting-people.org/archives/interesting-people/