‘The time for just talking is over. After years of promise, the age
of the videophone has finally arrived,’ says British Telecom. The idea of
a telephone that displays pictures as well as relaying sound is as old as
science fiction. But the electronics and telecommunications industries believe
that 1993 could be the year when people stop travelling and start using
videophones. Everything is in place: the technology to make the videophone
and the international standards to allow videophones around the world to
talk to each other. And the costs have now reached the point where it is
cheaper for companies to use videophone equipment than to pay for their
executives to travel.
Much as the British postal strikes of the 1980s opened the eyes of small
businesses to the benefits of using the fax instead of the mail, so the
Gulf War created a fear of flying that pushed companies into rethinking
their priorities on travel and conferencing. At the same time, a jigsaw
of new developments in digital technology were changing the rules of the
game.
But for people to change the habits of a lifetime and use videophones
instead of travelling to a meeting for face-to-face discussion, the firms
selling the videophones must present potential customers with a clear concept.
Several announcements appear every week, but most are written in a new jargon,
with no distinction between video conferencing systems, analogue and digital
videophones, and personal computers which double as videophones. Worse still,
the new videophone that British Telecom is launching this month, aimed at
the mass market, could convince the public that videophone technology has
a long way to go before it will benefit the serious user.
Advertisement
Conventional telephone lines are designed to carry analogue sound signals
and have a frequency bandwidth of around 4 kilohertz. An analogue TV signal
needs a bandwidth of at least 4 megahertz for clear moving pictures – equivalent
to a thousand phone lines working in parallel. Clearly this is impractical,
and calls would cost at least £50 a minute. The answer, until recently,
was to use runs of high-quality cable or satellite channels with a wide
bandwidth. Large companies have used such video links between sites in different
cities or countries for at least 10 years to set up video conferences. But
the hardware is complex, bulky and costs more than £50 000. It is
usually set up in a special studio or building, and the links have to be
booked in advance and cost at least £1000 an hour. There are still
only 5000 hardware units in use in the world, and different systems use
different, and incompatible, technologies. Ad hoc conferencing is all but
impossible, and in most cases, it is cheaper and easier for executives to
fly to a meeting.
The first videophones to use conventional 4-kilohertz telephone lines
went on sale in the US and Japan about five years ago. Aimed at the domestic
market, they work on the ‘slow scan’ principle to send pictures, of a sort,
down the phone lines. The videophone consists of a telephone, a black-and-white
screen about 15 centimetres wide and a video camera. The user dials the
number and, when a button is pushed, the camera takes a video snapshot of
whoever is using the phone, and converts it into digital code. The code
modulates a 2-kilohertz audio tone and the resulting warble, much like a
fax signal, is sent down the phone line at a rate of 8 kilobits per second
(kbit/s). At the receiving end, the incoming warble is converted back into
digital code, and stored in a solid-state memory. The stored code is then
displayed as a still picture on the recipient’s screen.
It takes about 5 seconds of warble to send the still picture down the
line, during which time the callers cannot speak. Picture quality is poor
because the digital processor breaks down the picture into 16 000 separate
picture points or pixels, compared with about half a million pixels used
in TV broadcasting. The kind of images sent could just as well, and just
as cheaply, have been sent by fax. And communication is only possible between
telephone subscribers who have invested in matching equipment. Prices initially
were about £200 per station, but soon fell to £200 a pair because
there was so little interest. The equipment is now to be seen remaindered
in North American mail order catalogues.
Undeterred, British Telecom is launching its new low-cost videophone
at the Ideal Home Exhibition on 17 March. Called Relate 2000, it plugs into
a normal analogue telephone socket and has a colour camera and a 7.5-centimetre
liquid-crystal display (LCD) screen for displaying pictures. It converts
about five low-definition pictures a second into digital code and sends
them down an analogue telephone line by modulating an audio tone at 14.4
kbit/s. The recipient’s unit converts the warble back into digital code
and displays moving pictures on the screen. The units, made by GEC Marconi,
cost £399 each.
Many people in the electronics and telecommunications industries wonder
how BT can afford to sell a colour videophone at this price. In the US,
AT&T has launched a broadly similar videophone which it sells for $1500.
Dealers found it hard to take BT’s videophone seriously when it announced
Relate at the annual round of consumer electronics trade shows in London
early last year. It could show only a prototype with a hole where the screen
should have been. But BT pressed ahead, frequently promising that Relate
would be on sale in time for Christmas. By Christmas it had managed to put
demonstration videophones into just two of its shops. A second launch date,
in February, passed by. Now, finally, BT has units ready for purchase by
members of the public wanting to install one of these widely-publicised
videophones in their own homes. Whether they will want to, however, is another
matter.
Conventional phone lines deliver such poor quality pictures that they
are acceptable only on a very small screen, which limits their use to ‘mug
shot’ dialogue. Moreover, like the AT&T videophone, Relate transmits
only five pictures a second. So any movement will appear as a series of
jerks and blurs on the recipient’s screen (TV sets display 25 or 30 pictures
a second to avoid this). Finally, the speech will sound unnatural. To travel
with pictures, speech has to be digitally processed but conventional telephone
lines cannot carry enough data to handle the information needed for good
quality sound and colour pictures.
Critics say that Relate is a compromise which could set back the start
of the videophone age by years. Even within BT, there are fears that when
customers see the quality of pictures that Relate delivers, they will write
off all videophone technology as a gimmick or toy. This will make it more
difficult for everyone, including BT, to sell more serious products.
The picture and sound quality available from a videophone system depends
on the amount of data that flows between transmitter and receiver, and this
governs the price of the hardware and cost of the call. The better the picture
quality, the larger the screen can be, and the larger the group of people
able to use the phone at the same time for a conference. The new generation
of digital videophones on offer, or coming soon, can offer a hierarchy of
performance options. Unlike Relate, these machines take advantage of two
key developments in digital technology which increase the amount of data
that telephone lines can carry.
The first is the Integrated Services Digital Network, which uses high
quality cables to link into the BT and Mercury networks. ISDN lines can
carry digital pulses at a data rate of 64 kbit/s, at least four times faster
than analogue lines, and with less risk of data errors. The digital pulses
can represent speech, text or pictures, which makes digital lines ideal
for carrying videophone signals. Using ISDN, users can dial video calls,
just as they dial fax calls.
Although countries have adopted different ISDN standards, most industrial
nations have built electronic gateways which let ISDN signals of one national
standard connect with local equipment. Since 1991, 12 000 businesses in
Britain have installed ISDN lines, and the number is increasing rapidly.
BT charges £400 for installing the line, and then £336 a year
for its rental. Within the next few years, these digital lines should be
available to domestic subscribers.
Moreover, ISDN lines can offer transmission rates of up to 384 kbit/s.
An ISDN line is laid as a pair of wires which carries signals in both directions,
for transmission and for receiving. Each wire pair can carry two data signals
at 64 kbit/s interleaved with each other, plus a third for control signals
at 16 kbit/s. The two data streams can be used separately or together to
give a capacity of either 64 kbit/s or 128 kbit/s. Lines can also be ‘ganged’
together to give a group of up to six data streams with a capacity of 384
kbit/s. For higher capacities, companies must install their own dedicated
digital lines, carrying pulses at 2 Mbit/s from their building to the national
telephone network.
However, even a digital line is not enough to carry the data generated
when the analogue signal from a video camera is converted directly into
digital code. Direct conversion from a good quality camera to code gives
216 Mbit/s. For transmitting TV pictures, broadcasters halve this to about
100 Mbit/s by digitally coding the TV signal after it has been processed
for transmission instead of as it leaves the camera. But this is still many
times in excess of anything a digital telephone line can carry. Data compression
is therefore the key to making videophones work.
Over the past decade there have been breakthroughs in compression which
allow the data rate of a digital video signal to be drastically reduced
without reducing picture quality too noticeably. Also, there is now an
agreed standard so that calls made from anywhere in the world can be taken
anywhere in the world. Standard H261, introduced over the past
year by the International Telegraph and Telephone Consultative Committee
(CCITT), covers signals at any bit rate from 64 kbit/s to 2 Mbit/s.
The CCITT compared all available compression systems, looking for one
that would work in real time, taking live video in at one end and putting
it out at the other, without introducing processing delays which would create
awkward gaps in the conversation. In H261, the CCITT has combined
the best aspects of several technologies, including discrete cosine transform
coding (DCT), differential pulse code modulation (DPCM) and motion compensation.
DCT breaks a picture down into separate block areas, analyses their content
and pinpoints differences, while DPCM describes these differences, using
the minimum number of digital bits. Motion compensation takes advantage
of the fact that when an object moves, its shape does not change much; therefore,
only its direction of motion and small changes in shape need be coded. Taken
together these processes let the codec (coder-decoder) reduce the overall
bit rate from more than 100 Mbit/s to any rate between 64 kbit/s and 2 Mbit/s.
The penalty is circuit complexity. Analysis, search and comparison in real-time
needs complex microchips processing 1700 million instructions per second.
At 2 Mbit/s, the H261 system codes and decodes at 30 pictures
a second to give smooth motion. With 64 kbit/s, coding and decoding is at
10 or 15 pictures a second. At half the speed of normal TV picture, the
viewer will find any motion jerky and see a smear, rather like a comet tail,
trailing behind the object or person moving. Between these extremes, the
picture quality varies depending on both the data rate available and the
amount of detail in the pictures.
In the H261 standard, motion compensation is optional in
the encoder (where it is expensive) but mandatory in the decoder (where
it is cheaper). So a low-cost videophone without compensation can still
take calls from a more sophisticated one with compensation. Speed is not
a limiting factor either. Like a fax machine, the codec sends handshake
signals down the line to tell the codec of the other phone the maximum speed
at which it can accept data. The remote codec then steps down its own rate
to match.
The CCITT’s intention is to ensure that any videophone built to the
H261 standard, anywhere in the world, will work with any other
H261 unit, anywhere else. But how this theory translates in practice
will only become evident as videophones go widely on sale. However, the
CCITT standard cleverly covers cameras and TV screens working in either
of the two major TV standards, PAL for Europe and Australia (with 625-line
picture, and 25 pictures per second) or NTSC for the US and Japan (525 lines
at 30 pictures per second). At each end of the video link, local standard
(PAL or NTSC) cameras deliver signals which are converted into a defined
Common Intermediate Format before transmission. At the other end of the
link, the common format signal is converted to the appropriate local standard
(PAL or NTSC) for screen display.
In Europe, telecommunications companies in Britain, Germany, France,
Italy, the Netherlands and Norway agreed in June 1991 to establish a pan-European
videophone service using the H261 standard. They will cooperate
to iron out any inconsistencies and to ensure a smooth transfer of calls
through the international gateways. Trials began in October last year and
the service should go live this summer.
In general, digital videophones which work at 64 kbit/s are acceptable
only for screens of less than 15 centimetres wide. The pictures are quite
blurred, and they jerk and smear badly if there is any motion in the scene.
The best results are obtained with images that remain static, because then
the memory chips in the phone have time to build up detail from the incoming
signal. In Britain, BT sells a personal digital videophone working at 64
kbit/s for around £3000. Introduced last year, it is very much a
personal desktop machine for the individual user.
More attractive to the business user could be the combined personal
computer and videophone system. BT, IBM and chip manufacturer Motorola
are collaborating on a personal computer that blends videophone images into
a multimedia display. The object is to let people talk to each other, across
the city, country or world, while seeing each other at their office desks.
At the same time they can exchange text documents. The PC displays the videophone
image on a window in its screen, alongside text or graphics. When displayed
in a window occupying about one-sixth of the PC screen, the 64 kbit/s picture
looks more acceptable. The cost is low. BT and IBM, using Motorola chips,
promise a package consisting of a PC with video camera, circuit board and
multimedia control software for some £4000. The equipment should
be available at the end of this year.
For most office desktop and small group calls, both data streams of
the ISDN line are used to give a data rate of 128 kbit/s. This gives sharper
pictures, less smear on motion and clearer sound. Just before Christmas,
BT launched its VC7000 conference unit (made by Tandberg in Norway). This
is a low-cost digital videophone, with a 25-centimetre colour cathode-ray
tube screen which is switchable between 64 and 128 kbit/s working. It costs
£7500. BT puts the cost of an hour’s video call at 128 kbit/s to the
US at £168, but Mercury promises to undercut this by nearly £40.
Call cost can be halved by switching the phone to work at 64 kbit/s, but
most serious users prefer to pay more to get the better pictures and clearer
sound available from 128 kbit/s working. The British telecommunications
giant GPT is also selling 64-128 kbit/s systems, with screens measuring
up to 70 centimetres, for around £40 000.
Centre of attention
Businesses are now starting to use the technology. Accountants Price
Waterhouse recently signed for delivery of six videophones to allow its
staff to meet international clients at short notice, without leaving their
office. The Ford motor company already uses video conferencing to resolve
production difficulties between its European factories, and National Power
uses desktop videophones to communicate between operational centres around
Britain. Tony Stewart, customer services manager at NP, says: ‘It’s more
personal. You get the other person’s full attention and build relationships
more quickly. You can see the reaction to what you say and see if someone
is paying attention.’
For large groups, needing larger screens, a data rate of 384 kbit/s
is necessary. Here, too, the price of equipment is falling. In October 1991,
BT introduced the VC5000 384 kbit/s system, which has a 70-centimetre
colour cathode-ray tube screen. Last November, it cut the price from £43
000 to £30 000. Line rental and line usage costs are higher than
the VC7000, however, because the system needs three ISDN pairs and clocks
up call costs at six times the price of a conventional call. But picture
quality is good enough for large screen display, provided that the conference
participants do not wave their arms about too much or run round the office
while talking. Anyone interested in seeing how a 384 kbit/s conference link
performs will have the chance to see one working on their own home TV screens
later this year, during the Whitbread Round the World Race .
The highest data rate that is used for video conferencing is 2 Mbit/s,
right at the top end of the H261 standard. Higher rates are thought
necessary only for broadcasting. The 2 Mbit/s rate cannot be handled by
ISDN lines and relies on one of the dedicated data lines used by large offices.
It gives picture quality close to broadcast standards and can be used where
the images must be displayed on very large screens, for instance to let
company moguls address staff or press gathered across the world in large
rooms or halls. Usually the participants will hire equipment or an equipped
studio. BT charges around £1500 an hour for conference calls made
to the US from a BT studio. Mercury has a studio in London, which gives
up to 24 people the chance to confer with the US for £1000 an hour.
At 2 Mbit/s prices, and with most conferences likely to last for more
than an hour, there is still an incentive to fly instead of use video. But
most group discussions can now be handled by 128 kbit/s electronics. At
these data and call rates it is usually cheaper to stay in the office. This
leaves perhaps the biggest challenge of all, to convince executives that
they must stay in a cold winter’s office instead of travelling first class
to a hotel with a large golf course set in a sunny clime.
* * *
COMPRESSION ON THE OCEAN WAVES
Competitors in the 1993/94 Whitbread Round the World Race will be using
384 kbit/s video conference technology to send pictures around the world
by satellite. When the yachts leave Southampton on 25 September, they will
be carrying video cameras and modified videophone equipment that will let
them beam pictures to TV stations for broadcasting.
The signals are sent by satellite link, using the network of satellites
operated by the International Maritime Satellite Organization, Inmarsat.
These satellites are designed to carry digitised speech and data at a rate
of 64 kilobit/second to and from small dishes mounted on boats and other
mobile installations. Although this rate is adequate for some uses of desktop
videophones, it is nowhere near good enough for TV news. But higher data
rates would need large dishes that are fixed firmly to the ground rather
than bobbing around on the ocean.
British Telecom has developed a ‘store and forward’ system which plays
a clever trick to improve picture quality. Crews shoot video pictures in
two-minute segments. The analogue signals from the video tape are then converted
into digital code, and the code compressed to a data rate of 384 kbit/s.
The two minutes of digital code is then fed into a solid-state memory, and
read out again at 64 kbit/s over a 12-minute period, and transmitted by
satellite.
News stations receive the pictures over 12 minutes, store them and
then play them back at six times normal speed, to reconstitute the original
384 kbit/s data stream. This is then decoded into two minutes of analogue
video.
Picture quality is acceptable, but rapid movements result in smears
and jerks. Quality is greatly improved if the data rate is doubled to 768
kbit/s and the transmission time doubled to 24 minutes. But this doubles
the power consumption of the transmitter on board the yacht, which means
the crew would have to carry more fuel for the generators which provide
the power. So most yachts will settle for 12-minute transmissions, and hope
that TV audiences are sufficiently excited by the picture content to ignore
the defects in quality.