Tip Of The Iceberg>
Moore's Law says processor performance doubles roughly every 18
months, and Shannon's Law says the more bits you cram into a channel, the
less certain you can be that they'll all make it through. Put the two laws
together, and you get the promise of 3G: Spectrum isn't getting cheaper or
more abundant, but each generation of radios is better at getting more bits
through.
“We think it might up to double our voice capacity,” said Oliver Valente,
Sprint PCS vice president, technology and advanced-systems development.
“We get that through some of the more optimized power-control
methodologies.”
Power control is critical for all CDMA-based 3G systems because it affects
the noise floor, which in turn governs the number of users the network can
accommodate. Leveraging as much of the 2G infrastructure as possible helps
reduce costs, and in many cases, 2.5G relies heavily on software and
hardware upgrades to existing elements. The HLR, for example, would be
upgraded to add fields such as whether a subscriber is data-capable.
Sites could be a different story. All of the 3G systems will need the same or
better link budgets compared to their 2G counterparts, but in order to offer
higher data rates, cells likely would have to be split. Even if the initial take
rate for the new services isn't dramatic, it still should be easier in the long
run to upgrade all sites rather than just a select few.
“If you didn't follow your frequency plan, getting more capacity becomes
more complex,” said Brad Fink, Nortel Networks senior product manager,
TDMA data.
Higher data rates put a burden on handsets, too. One challenge is
developing a reasonably priced, battery-efficient processor capable of
juggling several tasks at once.
“One area that (has) handset designers scratching their heads is how to get
the data in and out of the radio link when they're still having to keep up with
all the real-time processing required to maintain a wireless connection,” said
John Diehl, PrairieComm CEO & president.
That's important as data moves from a tacked-on afterthought to an integral
function. Here's one scenario: Suppose that a user has a smart phone that
uses IS-136 for voice and EDGE for data. When he isn't making a call, the
phone camps on an EDGE channel, where it's always ready to receive data.
“An incoming voice call goes through the existing infrastructure,” Fink said.
“The page is tunneled through these new network elements and to the site's
EDGE radio. The mobile is instructed that a call (is) coming in, so go to the
circuit network and get this call. When the call terminates, it goes back to the
data network.”
Beginning with GPRS, data traffic is offloaded to a new subnetwork of
data-only nodes. As in 2G, the base station controller (BSC) has a
connection to the MSC but adds a second connection to the data gateway
node. When the subscriber makes a call, the BSC makes a decision.
“If it's a voice call, it makes a connection to the MSC just like a normal GSM
call,” said Dan Bantukul, Tekelec IN Diagnostics division senior manager of
product marketing. “If it's a packet-data call, then the data gets routed to the
GPRS gateway node.”
Although that sounds deceptively easy, it involves more than simply
lashing together disparate elements, and vendors acknowledge that carriers
won't settle for high-maintenance, roll-your-own solutions. Manually
provisioning the connections is one option, but there might be a better way.
“You (could) provide some tools that essentially allow these nodes to
auto-discover one another and suggest how they think they should be
configured in a default mode,” Fink said. “All the baseline (settings) come
up on their own, but the operator can optimize things. That's what we're
striving for. It eliminates a lot of errors coming from the human side.”
New Freedoms — & Responsibilities
The RF network has its own set of new freedoms and responsibilities. GPRS,
for example, can vary its data rate automatically according to channel
conditions: As the subscriber moves toward the edge of a cell, and the
signal degrades, the network could decrease the data rate. Then, if the
subscriber is handed to another cell, and the signal improves, the network
would increase the rate.
This approach makes efficient use of network capacity because it reduces
the amount of packets that have to be re-sent to replace those lost because
of poor channel conditions. The downside is that optimizing networks with
multiple channel-coding and modulation schemes will be more involved, if
not downright tricky. Assessing bit-error rate (BER) is one example.
“At a given (location), the test equipment has to measure the BER for a
stronger coding, such as ¼- or 1/6-rate,” said Kamran Etemad, Wireless
Facilities senior manager, advanced-technology group. “At the same time, it
needs to measure the BER on a separate channel for a ½ or ¾. You need to
know what the BER is at that location using the stronger code versus the
BER for the weaker code.”
Optimization could involve determining acceptable BERs, which in turn
would help determine the optimum rate to make the most efficient use of the
available bandwidth while achieving quality-of-service targets. It also
requires upgrading T&M equipment.
“The test equipment may need to have multiple receivers, each operating at
a different coding rate, and each one measuring the BER separately,” Etemad
said. “You would know for this particular site whether you can have ¼-rate
or ½-rate coding all the way to the cell edge. Maybe halfway through the
site, you need to switch from ½ to ¼.”
People also will have to adapt to new issues and new ways of doing things.
Power control is one example.
“All of the GSM guys are going to have to be able to deal with that for the
first time,” said Ross Nelson, Textronix worldwide-business-development
manager, communications-test products. “One of their No. 1 concerns today
is learning how to do power management in the W-CDMA world.”
That's just one example of the learning curve that begins a steep climb at
2.5G. An early start certainly can't hurt. Sprint PCS, for example, continues to
trial vendor prototypes at its 3G test lab and appears undaunted by the
potential complexity.
“There will be some new power-control algorithms that we'll monitor, but it
wouldn't be substantially different from what we have today,” said Sprint
PCS' Valente. “It will still be forward- and reverse-channel power-control
parameters.”
Software-defined radios could provide a convenient, cost-effective way for
vendors and carriers to accommodate evolving, increasingly complex
standards, especially as competition forces new technologies off of drawing
boards and into networks much more quickly.
“That means you're not going to catch all the issues on the first go-'round,”
said Stephen Blust, Software Defined Radio Forum chairman and BellSouth
Cellular director of technology strategy and standards. “There's no way to
find the little gotchas until you've fielded systems, and maybe you've got
tens of thousands of subscribers on them. That's where software radio may
have its greatest payoff: being able to go back and in situ take care of those
issues without having to worry about board swaps and the things that
we've traditionally had to do.”
Comments? Write to tim_kridel@intertec.com.
wirelessreview.com
Ruff |
