Multimedia has different meanings for different sectors. For the entertainment industry it means new delivery channels (satellite, cable, video games) through which product can be sold or licensed. As far as multimedia is concerned, the PC industry seems satisfied, for now, with CD-ROM-based applications which feature interactivity, sound and animation. And in local area networks, multimedia hubs are those which connect different network technologies.
These definitions are not as disconnected as they at first appear: they all define an industry-specific view of one small piece of the same process. Digital technologies are now breaking down the barriers within which individual business sectors have been built. And once something has become digital it leaves the proprietary analogue world and becomes promiscuously connectable to every other digital device. So TV screens can appear on PCs, film channels can travel down telephone lines, cable TV networks can handle telephone access, and compact music discs can be adapted into computer storage devices.
From a technology user's perspective - whether a business or an individual - multimedia is a trend which allows different information, communications or entertainment media to become available through a single system or user device. The formerly separate disciplines of creating text, video and sound will then be able to be combined within a single product.
From both the provider and the user end of the telescope, multimedia seems to threaten - or promise - a revolution in the way information and communications are shared.
But although business leaders and industry gurus are agreed on the process, no one can be sure of the outcome.
Will television disappear, or will the TV set be the integrating device? Will the PC absorb television? Or will the new media be experienced through a completely new type of device? Will telecoms companies become major players in the programming business, or will a layer of intermediaries emerge? And that's just the business framework - what of the applications themselves?
To all of us on the receiving side of the screen, multimedia promises to end the constraints which artificially segregated the way we have traditionally received information on-line. The most efficient way of communicating with humans is to engage all of their senses and to make the communication interactive - the way we do it when we physically meet. In this sense, multimedia is not a gimmick - it takes us to the final destination of man-made media.
For the user, the hope is that the new capabilities will lead to the development of a rich medium offering all the advantages of animation mixed with text and sound (and perhaps even smell and taste, eventually), tied to an ability to interact in complex and individual ways with the source material. The challenge is to learn how to exploit it.
For the industry sectors involved, multimedia threatens to overturn existing businesses and business practices. The end result and the timescales within which the changes will take place are vague, but the enormity of the challenges and opportunities are readily accepted.
Interactive CD-ROM applications provide a starting point for the development of a specifically multimedia set of conventions through which users are able to interact with digital material.
This is behaviour and expectations at work, rather than the limitations of technology. As a trial of a multimedia banking kiosk in the UK showed, people can take time to adjust. The kiosk offered the usual automated teller machine alongside a two-way video link to bank personnel. But users had become so accustomed to the dumb anonymity of automated teller machines that they found the immediacy of a two-way video link disconcerting. The interactive link stayed, but the moving picture of the banking advisor was replaced with a still. Just because it's technically feasible doesn't mean it adds value.
After content, the first thing multimedia requires is bandwidth. PC multimedia is therefore based around the compact laser disc, or CD-ROM, which has the capacity to store large amounts of graphics. CD-ROM products also use MPEG compression to accommodate video sequences.
PC multimedia is already a substantial and fast-growing market. The CDs themselves are cheap to manufacture (now below US$ 1, in volume) and have the advantage of already being a well-understood consumer item.
Inevitably, although the physical medium is standard, the multimedia systems are not. At present multimedia users are faced by competing standards, although this endemic problem is to some extent alleviated by the programming in all of the systems. Many manufacturers now include software drivers for both Windows and Macintosh systems on the same disk.
The digital material is then accessed by, and displayed on, a relatively high-powered PC.
Multimedia products are ideal for games and educational packages, but although CD-ROMs are very powerful, they are best suited to self-contained applications and CD-ROM jukebox-type applications. One of the most popular products of this type is Encarta, a CD-ROM-based encyclopedia which takes particular advantage of the computer's dual abilities to search and to interact with the user.
The key to the development of multimedia was mass storage, and in particular the CD-ROM. Because high-density storage has arrived before affordable high-bandwidth, telecom services, applications are limited by the requirement to have the data source coupled directly to the computer, either locally or across a LAN.
But there are several reasons - including interactivity, ease of updating material, and economics - which suggest that ultimately data will reside remote from the user, and the public network will be able to deliver it.
Fibre optic networks are key, here. Fibre can deliver vast amounts of data, but new network switching equipment designs are needed because the existing time division channel switching methods are not suited for multi-gigabit fibre speeds.
Asynchronous Transfer Mode (ATM) is ideally suited to switching data across a network of high-speed circuits, and is also capable of carrying different types of traffic - computer, voice and video - without compromising their specific transmission requirements.
ATM has now been adopted by the general IT industry as the basis for the next generation of high speed networks, both public and private.
Increased bandwidth, and the remote storage and management of data, is the outcome favoured by the network operators, who see multimedia as a broadband network application, rather than as a computer application requiring increased public network bandwidth.
But there are signs that this is not an either/or choice. It is possible to envisage a scenario which uses a hybrid of the two.
Some of the more interesting developments on the Internet involve applications which couple CD-based high-density storage and real-time interactivity across the net. Rather than attempting to compress and download visual images, the local CD houses a library of visual objects and sounds which can then be assembled and modified by data running across the network.
The most obvious application of this technique is in the support of interactive, multi-user games, which make use of complex, animated graphics, so that the players are able to operate within a three dimensional space. But there is no reason why the technique cannot be refined for other applications.
Clearly, multimedia developers and users will design the applications to match existing cost structures - in the hybrid CD/Internet example, as high bandwidth services fall in cost and link speeds to the Internet gradually rise, so the interactive part of the application can take on a greater role in delivering the data.
Given today's limitations, it's hardly surprising that network-based multimedia applications are currently limited to primitive text, graphic and animation mixes, at the WorldWide Web end of the sector, or to high-value applications that can justify the hefty cost of bandwidth, at the other.
At the high end is Telesurgery, where high-quality video allows surgery to be orchestrated by a remote specialist. Further down the scale we are beginning to see the emergence ofspecialized commercial business services which can deliver text and video programming, a mix of which can be displayed on screen under the control of the user. But perhaps the leading high-value sector involves systems which combine teleconferencing and groupware, where participants are able to both communicate face-to-face and share files and data.
The development of multimedia will be influenced both by enabling technologies and by the slow growth in what will eventually be seen as primitive applications. As these are more widely used a set of conventions will emerge upon which the future, network-based multimedia will develop.
Asynchronous Transfer Mode (ATM) was originally developed by the telecoms industry as a method of switching the broadband services which could be made available by the development of fibre optic technology.
ATM is now being promoted as the basis of a unified multimedia network capable of delivering any type of data across the office or around the world.
BACKGROUND
The public network has traditionally dealt in electrical circuits or channels - fixed paths set up across the network for the duration of a call. With the introduction of first generation digital technology this fixed channel approach was maintained - each call being assigned a 64 Kbps path across the network - enough data to encode a good-quality speech pattern.
These digital networks use a time division method of organizing the channels. Multiplexers supervise the trunk circuits in the network by amalgamating dozens of individual 64 Kbps channels onto a faster circuit. A 2 Mbps circuit can accommodate 32 64 Kbps channels by assigning 32 rotating timeslots - all channels send just one bit in turn and when they are de-multiplexed at either end they arrive at the correct interval and in the right sequence.
The exchange switches use a similar time division technique to switch the channels to appropriate trunk or subscriber circuits. This approach works well for voice conversations: it introduces only minimal delay in the time taken for the data to travel from end to end; and it guarantees the fixed 64 Kbps capacity required by a voice conversation for the duration of a call. But it has proved to be a clumsy and restrictive architecture when applied to data or video transmission.
Computer communication, for instance, generally thrives under almost opposite conditions to voice. Instead of generating a steady stream of bits, computer to computer applications are bursty - suddenly requiring several Mbps of capacity and then lapsing into near silence for long intervals. Also, unlike voice, computer applications can be designed to be relatively tolerant of delay. Clearly, as the proportion of data traffic increases on the network, an architecture designed to accommodate both voice and various kinds of data (video, for instance) would be advantageous.
ATM solves all these constraints in a single bound, because it can control both delay and channel size dynamically.
To do this it diverges from current circuit switching technology in two important respects. Instead of maintaining fixed-capacity circuits across the network, ATM groups data into fixed length cells of 53 bytes (or Octets in telecoms jargon). Five bytes of this are reserved as a header to describe the data and its destination. The cells from one application multiplex themselves with cells from dozens, perhaps hundreds, of others, as they all cross the underlying, high-speed transmission circuits which make up the network. This asynchronous transmission method creates virtual channels with flexible bandwidth - the virtual channel becomes faster because the application is exchanging more cells, not because a fixed amount of capacity has been reserved for it. So the computer communications requirement to provide flexible bandwidth for bursty applications is supported.
But putting data into cells doesn't mean that unacceptable levels of delay need be introduced. Because ATM is designed for speed, an entire cell will occupy a broadband circuit for about the same time as a few bits will occupy a standard 64 Kbps circuit. In theory then, a cell carrying voice data can traverse a network without introducing a noticeable level of delay.
The second main departure from standard circuit switching involves the switching method. Because each cell can cross the network independently, the vital switching function can be made modular. A first-generation time division switch is essentially a high-speed computer dividing its attention between dozens of ports and switching the data between them. At each switching point throughput is limited by the speed of the switch.
ATM, on the other hand, can take advantage of a space division switch architecture. Because each cell knows where it's going, a matrix of simple high-speed switches can guide each cell, step-by-step, through a switching fabric - the greater the number of ports being switched, the larger the fabric.
In a switching fabric the logic is implemented in the hardware. Prior to entering the switch, a subsystem monitoring each circuit decides what output port the cell should be sent to by inspecting its header. It then adds a temporary routing header to the front of the cell which is peeled off, digit by digit, as the cell passes through the fabric. If two cells are simultaneously headed for the same output port (as in the example shown), one must be buffered there as the other passes.
This is a banyan fabric - there are others which achieve the same result. The great strength of this approach is that cells from different input ports can be switched in parallel so the switch doesn't become a potential bottleneck in the network.
Circuit switching is organized like a railway which runs trains with a fixed number of carriages, regardless of the number of passengers travelling.
ATM introduces private transport - a high-speed motorway is made available, and passengers then form small groups and may travel in an unlimited number of cars. Like private road transport, however, ATM has a problem when too many cars decide to travel along the same stretch of motorway. Having decentralized the switching function, ATM networks have to win back a measure of control in order to prevent congestion.
An important element here is for operators to provide different classes of service, so that priority may be given to cells supporting applications with time-dependency, such as voice or video. Cells supporting applications without time-dependency can be buffered or discarded if congestion occurs.
ATM therefore forms a promising basis for the broadband services required for network-based multimedia applications. Network operators can evolve their networks towards a single ATM infrastructure able to support all the different types of traffic that applications are likely to require in the long term (say 15 to 20 years). Operators can use ATM to provide specific services (like videoconferencing or LAN interconnect), but the real potential involves the multimedia integration taking place on the desktop or in the living room.
Click Here to Return to Telecommunications Features and Press