|
The Future Of Hardware VS Software
April 9, 1997
More than ever, the electronics industry is being driven
by the consumer's need for instantaneous results. Whether we dial up our
Internet provider and want instant access to all that lies within, including
audio, graphics and video, or we punch in a number on our mobile phone
and demand an instant, clear connection from a satellite, we don't want
to wait for results. Our new intolerance for latency stems from years of
longing for computer and telecommunications products that actually respond
to complex requests in something approaching real-time. That day is coming
soon, and DSP-based "soft" solutions will deliver the news.
Manufacturers of consumer products are racing to keep
up with the relentless march toward real-time performance. With much gnashing
of teeth, the industry has come to realize that some of its tried-and-true
theories about product development are no longer viable. Most notably,
it is accepting the need to keep hardware changes to a minimum while updating
capabilities in software.
It's a giant step for the electronics industry, which
has traditionally incorporated changes in its products only by introducing
new ones. Consider, for example, how many modems you have owned since the
days of 300 bits/second. From now on, similar changes will be incorporated
in the same modem (or other device) you already own, simply by copying
files from a diskette, or more likely, downloading them from the manufacturer's
Web site.
Cost, the universal driving factor of the electronics
marketplace, has long dictated that in new markets with low volumes, performance
enhancements are best implemented with software, because hardware in small
quantities is generally more expensive. When volumes increase, a point
is reached beyond which hardware modifications are less expensive than
software because the components can be procured in huge quantities.
However, since volume production today is usually reached
almost immediately after a product is introduced, logic dictates that enhancements
be made in hardware from the beginning, forgoing the software solution.
Nevertheless, this is increasingly not the case, for several reasons:
- The market has rewritten the formula. With mask generation
running in the hundreds of thousands of dollars, engineering time in short
supply and time-to- market shrinking, it takes too long and costs too much
to implement upgrades, bug fixes and new features with hardware alone,
even when volumes are huge.
- Consumers need a less painful upgrade path. Although
users desire more responsive performance and more features, they cannot
be expected to constantly buy new equipment to obtain them, especially
when the new features and capabilities are appearing monthly.
- Software upgrades are extremely marketable. Manufacturers
of products such as modems are well-served by allowing buyers to upgrade
instantly by, for example, downloading software from the manufacturer's
Web site.
- Upgrades quickly reach the customer. New standards,
features and improvements can be obtained by the consumer in a fraction
of the time required by the hardware route.
- New control and revenue generators are available for
wireless-service providers. With the ability to activate or deactivate
a feature "over the air," service providers give their subscribers
the option to use only the features they desire. When the only cellular-telephone
option was call waiting, this was of little significance. But the new digital
personal communications services (PCS) offer a host of features, making
this a marketable commodity.
For these reasons, software is increasingly the chosen
vehicle for change, even at high production volumes, where hardware changes
were always the accepted technique. It is simply faster and much less expensive
for manufacturers to introduce new products and upgrade those in the field
with software rather than hardware.
Data communications standards and wireless PCS provide
good examples of how the market is erasing the hardware/software crossover
point. Just when modem manufacturers had V.32 down cold, along came V32.bis,
then V34 and V34+, and now x2. In the past, each time data communications
technology boosted speed to the next plateau, many modem manufacturers
implemented the changes with modified and additional hardware along with
software. But many V.34+ modems from various manufacturers will be software-upgradable
due to this change in product concept.
In the wireless example, communications protocols or "air
interface standards," such as IS-95 and IS-136, have come from birth
to relative maturity in only a few short years. Manufacturers of wireless
products have been tracking their progress through the technical and regulatory
maze in order to know which ones to support and how to design products
that incorporate them.
Multiple features
In addition, the phones need to support many features
such as alphanumeric paging, voice mail, call waiting, call forwarding,
Internet access and data communications capability (such as Cellular Digital
Packet Data). They must also operate with not just two, but ultimately
many air interface standards, all within a single, palm-size enclosure.
For example, if a phone handles full- rate and enhanced full-rate Global
System for Mobile Communications protocols today, it may also need to accommodate
IS-95, IS-136 or any of the other standards tomorrow. Finally, vocoders
in digital wireless phones must be replaced to provide higher-quality voice
services.
The manufacturer of a wireless PCS set that handles three
air interface standards has two choices: create three separate sections
within the phone, or one section that can be programmed to be any one of
the three by software. It's not hard to guess to which direction the industry
is turning.
Software programmability makes it possible to implement
these changes over the air by downloading code into a programmable, DSP-based
subscriber set. The subscriber can even choose to use the slower speed,
for a reduced per-minute air-time charge.
The major cellular and PCS phone manufacturers are already
putting more flash memory in their products because the phone must be programmable
to embrace what the future holds. In the early stages of programmable phones,
the changes will be performed at a service center. But over-the-air modification
will ultimately take over as the primary delivery method. File download
could be done at times when communication traffic is light.
As a class of equipment, modems are far ahead of wireless,
which is just now making the change to full DSP-based programmability.
In modem design, V.32.bis was the turning point from hardware to software
solutions. New communication standards like x2 must be added to the many
existing standards that a modem must handle, right down to Bell 103. Implementing
all of this in hardware, while maintaining the ability to accommodate future
enhancements, is almost unimaginable.
The architecture of some of the more advanced modems has
already changed to satisfy future needs. These products have remained almost
the same since V.32.bis, with only minor hardware changes. Enhancements
and new standards can be made in software, either via diskette or Web-site
download. This was initially done for the convenience of the manufacturer,
but is now available to the consumer.
Perhaps the most visible of these companies is U.S. Robotics,
which developed x2 and is already shipping products with 56-kbit/s download
capability. Since last year, the company has been shipping modems that
can be upgraded in place from V.34 to x2 without hardware modification.
The ability to do this stems from its extensive use of DSP, specifically
the TMS320C5X family of devices.
Another likely candidate for DSP programmability is the
television set-top converter, which soon will have to handle HDTV and Internet
browsers as well as traditional NTSC TV. Rather than provide all of this
capability in the set-top converter, only the required software need be
resident, the remainder being downloaded when needed. This makes the converter
less expensive to build and gives the cable company more control.Whether
the end product is a wireless phone or a modem, it interfaces with the
public communications network. As a result, it is possible to upgrade flash
or RAM-based software in place via this network.
Many choices
In the case of wireless devices, a product may be programmed
to simply receive stock quotes, to transmit and receive voice and data,
to download billing and personal information, or to do them all. Think
of it as a software platform rather than a device with a function predetermined
at the factory.
Looking out a bit further, it's possible to imagine a
wireless phone that looks for and determines the cost of the least expensive
service provider at the moment I make the call, and then logs on to that
system. The service provider identifies me and applies the features I have
indicated I wish to use. It then downloads those features to the phone.
As the service provider adds features to the network,
they are immediately available to me via over-the-air download, as are
a virtually limitless number of other capabilities. Some enhancements can
even be made transparently, much as Internet providers do today.
These are only a few obvious examples of where DSP-based
software solutions are likely to be first realized in commercial products.
But the potential applications are limitless. So in the future, perhaps
we'll just download the means to do what we want over the Internet in real-time.
Source: Electronic Engineering Times
|