*Very good wireless data article* Note the advantage for U.S. CdmaOne carriers over their tdma competition in the absence of additional 3G spectrum allocations.
By Ira Brodsky - WirelessReview wirelessreview.com
Anyone who has implemented wireless data knows actual through-put may
be considerably lower than advertised speed. However, it?s packet radio
rather than false advertising that is to blame. Protocol overhead, noise and
other users sharing the channel conspire to ensure no one ever
experiences anything close to 19.2kb/s performance on a 19.2kb/s packet
radio service.
But the wireless industry has been reasonably candid compared to the
dial-up modem industry. Many people have 56kb/s dial-up modems, but
have they ever actually connected at 56kb/s? That only happens when you
call a service using a special digital line and you have a clean line at your
end. At least with a packet radio service ? even if customers naively
believed they were running 19.2kb/s ? they would know instantly that
they couldn?t afford high throughput in anything but small doses: Most
packet radio services charge by the kilobyte.
The Other Half of the Story
The speed at which the bits move is only half of the story. What good
does it do to receive a short message at blinding speed after a 30-second
delay? Latency, rather than low bit rate, is what has tripped up wireless
data. And latency is especially pernicious in today?s IP-based world. IP
entails frequent back-and-forth data exchanges, so delay tends to pile up.
What sounds like moderate latency quickly can become intolerable.
Wireless engineers are fond of saying that most applications were
designed to run over wires and assume unlimited bandwidth. That isn?t
entirely true. It would be more accurate to say that most applications
assume latency on the order of 10 milliseconds ? roughly the time
required for signals to travel round trip over terrestrial communications
facilities. Unfortunately, data sent over today?s packet radio networks
incurs delays ranging from 100 milliseconds to several seconds.
The Bottom Line
Expect the average throughput of a packet radio service to be a fraction of
its nominal bit rate. Whether that fraction is 3/5 or just ¬ depends on both
network architecture and current traffic load. Some packet radio networks,
such as Metricom?s Ricochet, provide fairly good throughput (28.8kb/s)
because they operate at a much higher nominal bit rate. Others, such as the
Ardis and BellSouth Wireless Data networks, have sidestepped the issue
by repositioning themselves as 2-way messaging services ? a role in
which they are starting to have some success.
Wireless-data services recently announced by cellular and PCS carriers are
circuit-switched rather than packet-switched. Expect these services to
perform at their advertised speed: 14.4kb/s. Average throughput may be
lower when signal quality is marginal. On the other hand, average
throughput may be higher when signals are strong and content is
compressible.
Basically, this means you can count on these circuit-switched
wireless-data services to work with existing applications, doing pretty
much anything you can do with a dial-up modem. The primary exception is
that it may take a bit longer to download attached files, and mobile users
are often in a hurry, so it is important to have the option of skipping (and
saving) such messages ? or at least their attachments.
Is 14.4kb/s (circuit-switched) fast enough to finally kick-start the
wireless-data market?
The widespread deployment of 14.4kb/s, cellular- and PCS-based,
circuit-switched wireless-data services is a major development. It is enough
speed, if just barely, for exchanging e-mail and accessing most Web sites.
The Sprint PCS announcement is particularly important because it entails
11,000+ base stations nationwide; the data services can be accessed from
both standard handsets and smart phones; and Sprint PCS is collaborating
with key Internet players such as Yahoo! This announcement puts
pressure on other wireless carriers to move beyond the seemingly endless
trials and limited commercial deployments.
Never Enough
However, mobile users will demand and require even higher speeds. This is
a huge opportunity because dial-up modems have reached an evolutionary
dead end: You just can?t squeeze much more throughput out of a 3,300Hz
analog voice channel. The wireless industry, in contrast, has a relatively
clear path to developing 64kb/s services soon and 384kb/s services later.
At first, carriers will be able to charge higher rates for higher speeds. But to
grow the market, they will have to reduce rates gradually and increase
network capacity. This is a problem for wireless carriers because a single
64kb/s circuit-switched data connection uses as much capacity as several
voice calls. Carriers will need to develop more efficient ways of providing
high-speed services ? perhaps a hybrid circuit-switched/packet-switched
solution ? but, ultimately, they will need more capacity.
If U.S. wireless carriers are going to create a big wireless-data market, then
the United States will have to allocate additional spectrum for 3G and even
enhanced 2G services. Multimedia applications are going to consume more
bandwidth, and the number of people accessing the Internet and intranets
is going to increase. Sure, engineers will invent ways to squeeze more
information in fewer bits, but others are busy inventing 3D browsers,
virtual worlds and telepresence ? applications that will demand vast
bandwidth.
Lurking Competition
Some industry participants claim the United States doesn?t need new
spectrum for 3G. This is shortsighted if not self-defeating. The cdmaOne
industry could reap a temporary advantage from such a policy because it is
probably the only camp that can squeeze voice and high-speed data
services in a 30MHz PCS channel. The opportunity is to make wireless data
as common as wireless voice. That will require much more capacity.
Plus, cellular and PCS carriers soon will face competition. MobileStar is
planning to provide high-speed access in places frequented by business
travelers ? mainly airports and hotels ? employing Proxim?s RangeLAN2
1.6Mb/s wireless LAN technology. Although wireless LANs offer limited
coverage, several hundred strategically placed nodes could attract tens of
thousands of road warriors.
Fortunately for the mobile telephone industry, there are a couple of
underused bands close to the PCS band: MMDS and WCS. For now, most
MMDS licensees are focusing on fixed applications, but with as much as
200MHz per city, they surely would have room for 3G technology.
The new 14.4kb/s circuit-switched services may have encountered some
speed bumps, but they should prove just fast enough to get the market
rolling.
Brodsky (ibrodsky@ix.netcom.com) is Datacomm Research Company president. |
