THIS ARTICLE WAS SCANNED FROM BYTE MAGAZINE (JULY-92 ISSUE) 
Getting a Broader Picture

Broadband ISDN and HDTV look set to become
the multimedia transmission standards of the future

There are signs that ISDN's apotheosis is not going to
happen: ISDN is apparently failing to measure up to
the standards that will be required by multimedia
over the next few years. Telecommunications ex-
perts all over the world agree that B-ISDN (Broadband
ISDN) is the system of the future. Researchers like Vinton
G. Cerf, president of the American Corporation for Nation-
al Research Initiatives, says, "As components of the B-
ISDN system design, ATM [asynchronous transfer mode]
and optical transmission methods could gain the degree of
importance in the twenty-first century that the telephone
network has today." Cerf believes that in the future users
will send enormous volumes of multimedia radio, TV, and
video data by cable. He also believes that radio frequencies
will have to be reserved for mobile data communication via
PCs. This is plausible, but it also raises a question: Which
medium is preferable for transmitting images? Certainly not
ISDN.

B-ISDN and ATM

B-ISDN is the solution offered by private network providers
such as Siemens. Currently, TV-signal distribution between
TV stations or to terminals starts at a throughput of 33
MBps; transmitted by ISDN, the picture setup would be too
slow to be practical. Color telefax requires a bit rate of at
least 2 MBps, and ISDN is three times too slow even for
that. Similar figures apply to almost all applications within
the scope of broadband services (see the figure).

Professionals consider ATM to be the appropriate trans-
mission method for broadband services. While ISDN's
STM (synchronous transfer mode) combines the circuit
switching of telephony with the packet switching of data
communications, ATM packets the entire data volume in a
uniform way (see the table).

ATM doesn't allocate a specific channel to information;
instead, it packets the information in ATM cells and sends it
over a uniform network. The packetizing facility, combined
with the downstream multiplexer, packets information in
cells that each have a 48-byte information field and a S-byte
header. Only one transmission facility is needed for all data
types.

Multiple channels are available with ATM, and the speed
of the receiving terminal is irrelevant. You can freely select
transmission bit rates. Cells from different sources, with
different bit rates and contents (i.e., voice, video, and data),
can be transmitted on an ATM-oriented network without
other links. ATM is thus a good unifier, cushioning with its
flexibility the numerous migration steps from slow to fast
terminals and services. In the foreseeable future, a 135-MB
ATM-oriented B-ISDN will be able to transmit any kind of
information imaginable today, probably via optical wave-
guides.

The problem for computer manufacturers is that their
present terminals are unable to receive and process the large
volumes of data that are typically transmitted through B-
ISDN. Document sizes of up to 4 MB, the usual size today,
will seem ridiculously small compared to future documents, 
which may consist of several gigabytes.

Furthermore, for many users, compression is really not an
alternative. Employees working on pilot projects report that,
often, customers do not want compression. For example, when an
aircraft manufacturer holds a video conference with experts to
assess the condition of an aircraft engine, a portable camera is inserted into the engine, and the resulting pictures are trans-
mitted to the screens of the participants. Is it safe to start the
engine, or do the blue discolored patches indicate defective ma-
terial? Such questions can be answered only if the picture is
transmitted unadulteratedin other words, uncompressed.
Similar conditions apply to the transmission of x-ray images be-
tween hospitals. For reasons of law and precision, compression
can be undesirable for this task.

Using B-ISDN for these critical applications could lead to
enormous increases in productivity and savings. In the aircraft
example, every minute the aircraft in question is on the ground
costs the company money, and the specialist may be on the
other side of the world. Compared to the cost of getting the spe-
cialist to travel and look at the aircraft in person, high-defini-
tion video conferencing has many potential advantages.

Workstation Inadequacies

From the point of view of compressed high-volume data and
highly sophisticated user interfaces, today's workstations are
lacking in many respects. Even high-performance worksta-
tions, like the IBM RS/6000 Model 220 (which can operate at
25 .9 SPECmarks) and 486-based PCs, are looked upon as inad-
equate terrninals for use with multimedia applications.

The SO-MHz RISC processors, like the future PowerPC chip
from IBM and Motorola, will someday be at the low end of the
performance range. The RISC principle, however, is consid-
ered sufficiently expandable to meet future requirements.

Within today's machines, problems occur at all points where
data has to be transmitted. Experts consider an IBM Micro
Channel machine's performance (a theoretical 160 MBps) to
be insufficient for a workstation. Thus, they are thinking of
using ATM as a bus station even within workstations. A new
bus structure is therefore essential.

Sun Microsystems is currently developing ATM technology for bus and LAN use. As Bill Joy of Sun believes, users would
otherwise be forced to reintroduce a central computer room to
their operations, because the distances between workstations
could not be bridged fast enough. Current RAM chips are too
slow; so are hard drives. New technologies, like static RAMs,
are too expensive.

The RISC operating systems available today rarely have the
real-time capability necessary for serious multimedia applica-
tions. Professionals say that pictures can be manipulated in real
time. More appropriate operating systems for multimedia are
MACH from Carnegie Mellon University (which is used in the
Next system) and Chorus, which emanated from an Esprit proj-
ect. These operating systems are not yet able to orchestrate in-
coming data streams; video and sound have to be rendered in an
exactly synchronous way.

A further pressure on these new operating systems is the
principle of a microkernel architecture, where the kernel of the
operating system is very small and (in principle) hardware in-
dependent. The hardware-dependent components are clustered
around it so that it's possible to emulate other machine designs
on the kernel. Windows NT (New Technology) from Microsoft
was developed according to this prinaple.

Data Formats

The question of which data format to use is a tricky one. What's
appropriate for users in the professional field is different from
what's needed in the low-end sector. Many users in the latter
group will opt for Apple's QuickTime; for the former group
Open Document Architecture (ODA) will be extended by mul-
timedia components.

In the event that compression becomes necessary, profes-
sionals have agreed on the MPEG (Moving Pictures Experts
Group) standard. MPEG compresses at a ratio of 1 to 20, which
is relatively low, but it has the advantage that little information
detail is lost. This type of workstation will probably look like
the prototypes that have already been built: built-in loudspeak-
ers, a camera, a videophone, and a separate camera for still
pictures. It will probably have an internal ATM bus and an ex-
ternal ATM adapter to handle all communications functions.

Since ATM supports a LAN topology and has a uniform ad-
dressing scheme, workstation clusters, PABXes, PC servers,
and multimedia servers can be integrated in this way. Network
monitoring, still a problem, can at least be made more uniform.

One Monitor Only

The notion that there will soon be only one type of monitor puts
a lot of time and performance pressure on PC manufacturers.
Only a few computer companies, such as Apple, seem to appre-
ciate this fact, however. Companies in the consumer market, on
the other hand, are mainly concerned with communications.

Consumer-oriented companies are concentrating their future
plans on screen technology, as Sony is with Trinitron. These
companies are sparing no effort; in a few years' time, people
will have better screens on their TVs at home than they have on
their office computers. If the marketers of consumer products
are successful in this strategy, they will take a hefty share in
portions of the market that computer manufacturers currently
think they have safely in their pockets.

If this happens, HDTV will be the natural successor to XGA.
In April, Fujitsu, Hitachi, and Sony started shipping the first
commercial versions of their new HDTV decoder chip set,
which consists of 12 ICs (as compared to the 150 ICs in the
previous version). This will make it possible to build HDTV
devices at a considerably lower cost. If competing standards
had not evolved for single decoding (e.g., MUSE in Japan;
HDMAC/D2-MAC in Europe, and several standards in the
U.S.), HDTV would probably have shown its paces earlier. In-
dustry insiders now seem to regard HDTV as the standard that
will take over from XGA.

Whenever people talk about HDTV, the subject of high
transmission rates comes up. Standa~d HDTV screens massive-
ly outperforrn high-resolution color VGA screens, with 384 MB
instead of 24 MB per image being transmitted down the line.
HDTV distribution starts at a transmission rate of 135 MBps
210 times higher than anything in current use.

The Wave of the Future?

What is missing from the picture? In principle, only the appli-
cations software to conform to the new standards. Soon there
will be a demand for a software developer's kit that allows soft-
ware houses to write and test communications software without
having to use existing hardware. Such a kit would include an
ATM board, the corresponding API specifications (low- or
high-level), and a box that emulates an ATM switch.

Is it worthwhile to invest in multimedia at this point? Pilot
companies that are using it have great hopes for it. A number of
laboratory models have been builtand a lot of field tests have
been performedwith the new broadband networks and ser-
vices in Germany. Among them is the BERKOM (Berliner
Komrnunikations System) project, which implements an initia-
tive of the Berlin Senate and the German post office under the
project management of Detecon. In 1989, Siemens succeeded
in performing the first data exchange via ATM.

Before you invest in ISDN or multimedia, you should find out
about the BERKOM project, talk to Apple, and watch a demon-
stration of the capabilities of ATM technology. All this may
well convince you that ISDN is too little, too late. 

BIBLIOGRAPHY

Siemens AG. "The Intelligent Integrated Broadband Network
Telecommunications in the 1990s." Issued by Bereich Offentlich
Kommunikationsnetze, Zentrallaboratorium, Postfach 70 00 73,
D-8000 Munich 70, Germany.

Susanne Mller-Zantop is an industry analyst and publisher of
the German PC newsletter TidBits. She is interested in the suit-
ability of new PC technologiesfor users in general. You can con-
tact her on BIX as "susanmz. "
