AGOCG logo
Graphics Multimedia VR Visualization Contents
Training Reports Workshops Briefings Index
Also available in Acrobat format Back Next

CONTENTS


1 Introduction
2 What is Multimedia?
3 Pedagogy and technology

4 Networks
4.1 Presenting multimedia information
4.2 Networked systems
4.3 What is ISDN
4.4 Multicasting
4.5 The role of ATM
4.6 Videoconferencing

5 Future Work
Glossary
Appendices
Bibliography


Case Studies

Multimedia in the Teaching Space

4.6 VIDEOCONFERENCING

Video conferencing will be broadly divided into ISDN and IP techniques. The difference is essentially one of guaranteeing the bandwidth available, and therefore the reliability of the quality of the picture. When using ISDN the user is able to obtain from the Telecom service provider a guaranteed bandwidth on the network on demand. The quality of the pictures is consistent with that bandwidth and any deterioration is more likely to be related to other factors in the network operation or the equipment performance which is being used. When IP is used then the bandwidth available to the link will vary with the number of users at any particular time on the network, and this cannot be guaranteed. Further if a new user comes onto the network during the session there is increased competition for the bandwidth. Thus the variability of the bandwidth leads to variation in the quality of the picture. As video-links are real-time, any reduction in the bandwidth available results in less information coming over the link and the picture quality falls. On the whole the ability to transmit video on the UK academic network, SuperJANET, is not impossible, but once one wishes to connect to sites outside UK there are major difficulties. If European connections are made over JAMES or the TransEuropean Networks then the situation is better.

There are efforts being made to develop software tools which will guarantee a limited bandwidth over LANs and the INTERNET, and these will improve matters. The real solution is making the bandwidth available on the INTERNET much larger, but then the problem may be that the use will expand to fill the available bandwidth with no overall improvement, in a manner analogous to the way road traffic increases when better roads are provided. The position with IP link is that much work is being carried out to improve the service and this will continue into the future. The service currently available is probably tolerable for point-to-point operations and for individuals communicating, but the service is not sufficiently reliable for important teaching application in a general way.

Videoconferencing depends upon the analogue to digital conversion and compression of video and audio signals for their transmission between sites and then the decompression and digital to analogue conversion of the signals for their presentation at the remote site. There are several encoding/decoding protocols which are currently in common use (see section 2.7). There are a number of proprietary systems also in use and the user has to make a decision whether to use the accepted standards. If one is sure that always and forever you will only be communicating with equipment from a specific manufacturer then you can use their proprietary algorithms, but this will preclude communicating with the systems of another manufacturers; e.g. the standard algorithm for encoding video to be transmitted over ISDN is H.261 and all manufacturers implement this capability in their equipment, hence compatibility problems rarely exist currently between the equipment of the best known manufacturers.

Other encoding algorithms are MPEG-1 and MPEG-2. MPEG-1 was developed for CD-ROM applications and works in the region of 1-2Mbps. This produces video of a quality comparable with good domestic television, and at the upper level of the bandwidth range there is little delay in the movement in the picture. MPEG-1 visually is very similar to H.261. MPEG-1 systems are easily available and the encoding and decoding equipment is not expensive.

H.261 is normally used in ISDN transmissions and cover the range 128Kbps to 2Mbps. One finds that at the lower end of the range the picture is delayed and there is a severe lack of synchronisation between the movement of the lips of a person speaking and hearing the audio. As the bandwidth increases to 384Kbps this synchronisation problem largely disappears. This bandwidth corresponds to the ISDN-6 service, and the delay in movement is also very much reduced. These improvements will continue up to 768Kbps, but above that figure to the maximum of 2Mbps there is little perceptible advantage. The H.261 algorithm is mainly working on the white-noise levels of the original picture and this does not result in major improvements in the visual quality of the picture.

MPEG-1 and H.261 standards are averaging effects over several video frames and consequently the precision of control in interactive systems are more complicated.

Diagrammatic representation of the H.320 Standard

This diagram show all the individual standards that go to make the full videoconferencing standard H.320. This show that there are two fundamental streams for audio and video. The video compression algorithm is H.261, providing 2 resolutions, CIF and QCIF. In addition to these two sections, there is a communications channel and a Data-sharing channel. The communications channel deals with matters such as framing, establishing the communications links, supporting multi-point working and provision of control for cameras at the remote sites. The Data-sharing channel is concerned with the T.120 standard which supports collaborative data sharing and whiteboards.

Another algorithm is H.263 which is specially designed to improve picture quality at lower bandwidths and is applied in LAN systems. It is also used in some ISDN systems and does bring about a perceptible improvement in picture quality, see section 4.5.1.

MPEG-2 is a higher bandwidth algorithm for transmission rates above 2Mbps. The MPEG-2 standard was developed for cable systems where the encoding process required a powerful computer at the head-end of the distribution system, but the decoding was cheap and could be supplied easily in a large number of user systems without raising the costs significantly. The quality of image that can be achieved as very good but there are relatively few MPEG-2 systems in use routinely primarily because of the cost of encoding the video material.

Motion-JPEG is another algorithm which is used extensively in the Metropolitan Area Networks, e.g. it is used in the Scottish MANs and the quality of the video pictures are very high. This algorithm works primarily on each video frame in turn and with high processing power can process the video frame by frame in real-time. The high processing power means that the delay in the video is very small, but the equipment is more expensive.

4.6.1. The development of H.323 Standard for Videoconferencing:

In contrast to seeking high resolution images with high speed network, there is another field of development which concentrates on the H.263 video encoding algorithm which provides better images at the low bandwidth end of the spectrum. This algorithm was developed to encourage more efficient use in the region of 128Kbps and lower, and it also had applications on LANs in mind. Where the INTERNET was being used, and when the bandwidth had to be shared with other non-video applications it is important to have the best video possible in the restricted bandwidth.

The H.263 standard is discussed in section 2.7.2.2. H.263 is the coding algorithm underpinning the H.323 videoconferencing standard, and some people are of the opinion that in time it will replace H.320. H.323 is beginning to find its way into ISDN video conferencing systems and it produces a better picture at 128Kbps than H.320. However if it is applied at higher bandwidths the advantages are not so apparent.

This is a video coding standard and was published around 1995-6 , and it differs from H.261 in the following ways:-

It uses half-pixel precision for motion compensation where H.261 used full pixel precision. Some parts of the hierarchical structure of the data-stream are now optional, so that the CODEC can be configured for a lower bit-rate., or better error recovery.

There are four negotiable options to improve performance;
  1. Unrestricted Motion Vectors
  2. Syntax-based arithmetic coding
  3. Advanced prediction.
  4. Forward and backward frame prediction similar to MPEG.

H.263 is supposed to be capable of providing the same quality at half the bit rate to H.261, and also H.263 support 5 resolutions which enables it to compete with the MPEG standards.

4.6.2. Videoconferencing for Meetings.

The early experiment on the SuperJANET ATM Video network were with videoconferencing. The common concept was that all that was necessary was a camera from the local video shop and some loudspeakers as might be available with one of the PCs in the departments and all was ready. The experience of SuperJANET was that did not produce good results, the pictures were of poor definition, the audio was difficult to hear, and all kinds of things might happen, from unpleasant echo effects to howling as the microphones resonated.

The Audio-Visual departments in some of the principal SuperJANET sites in the first SuperJANET experiments were involved in setting up a teaching project and they were able to see that simply by implementing good audio-visual practice that simple remedies could be applied and these were very successful. Other problems like the malfunctioning of echo-cancellers were not so simple and it was decided to set up the Audio-Visual Consultancy based upon a collaboration of the Universities of Cambridge and Newcastle-upon Tyne with University College London.

This consultancy set in 1993 and it set about inspecting all the SuperJANET sites, to check there room design, the equipment being employed, and investigating how the audio-visual signals were being presented to the echo-canceller. Very quickly this became an investigation into the performance of the echo-cancellation equipment, and it was found that in a very significant number of cases the equipment was incorrectly installed. Once these investigations had been completed there was an important improvement in the quality of the pictures and audio on the network. There have been considerable improvement in equipment since those times, and present day equipment usually has competent echo-cancellation equipment incorporated. However if this equipment is to be sued in large room such as lecture theatres then often a free-standing echo-canceller is necessary. (Document about echo-cancellation in http://www.video.ja.net_

The lesson to be learned form this exercise was that the collaboration of the Network Groups and the Audio-Visual Groups was important to the successful running of the network and this collaboration has continued now for several years; it has led to the setting up of the Video-TAG, the publication of a number of papers on the WWW on aspects of video conferencing, the development of databases to help potential users about the equipment available, and the development of testing protocols for audio and video performance of videoconferencing equipment. This has now reached the stage that a national Videoconferencing advisory Service has been set up on 1st January 1998, run jointly by the University of Newcastle-upon-Tyne and University College London. Information about this advisory service is available at the URL: http://www.video.ja.net

4.6.3. Videoconferencing - Teaching versus Meetings:

One of the most important observation form this exercise was the realisation that videoconferencing for meetings was not the same as for teaching. This is important as increasingly people are looking to the possibilities of using video conferencing in teaching.

The following are a list of characteristics of a typical videoconferencing meeting:-

  1. Most of the people taking part are of a similar status, and so the interaction takes place between "equals"
  2. The session is usually required to be highly interactive.
  3. In many case the meeting is point-to-point, although the number of multi-point meetings is increasing.
  4. The meetings are often designed to meet a problem between people working at some distance apart and part of the reason for setting up the meeting has been to save on travel.

In teaching the participants have different roles.

  1. There will be a teacher and there will be pupils who usually will regard the teacher as the authoritative figure.
  2. The session will depend upon the teacher for the level of interaction, i.e. how well the interaction takes place depends upon how well the teacher stimulates the pupils and encourages them to interrupt with questions.
  3. The sessions are likely to be multi-point as the use of videoconferencing is associated with a form of distance learning and teaching, and the motivation for using videoconferencing will be the ability to reach a number of small groups of students who are distributed widely.
  4. The session are not usually designed to handle a specific problem but to deliver a series of tutorial/sessions over a period of time.

This list is by no means exhaustive but it give examples of the types of differences between teaching applications and meetings held on the network. There are other comparison which can be made between the use of videoconferencing in the universities and colleges over the ISDN network or the INTERNET. Currently many applications on the INTERNET involve desk-top systems, with a small personal video camera and small loudspeakers alongside the PC or UNIX machine. This type of system has developed a poor name because the same level of care has not been taken to ensure that the user works in good lighting conditions and so makes most efficient use of what is in fact a cheap video camera, and that the microphones are placed to avoid resonance with the loudspeakers and do not pick up extraneous noise from the surroundings. The quality of picture that can be obtained is directly related to the bandwidth available, and if similar bandwidths are available on the INTERNET as on ISDN the picture quality should be comparable.

At the present time the use of desk-top systems is attracting a lot of interest, and if the bandwidth available within campus networks is sufficient then one can expect increasing use. The issue is if too many people take up these facilities on the campus, then the network will suffer. People should remember that the video on these systems is often only providing 5-10 frames per second as opposed to conventional video which is 25-30 frames per second; and hence the images are jerky. It is the decision of the teacher whether that is acceptable for their purposes. In practice (and as recommended by the videoconferencing strategy document) a bandwidth of 384Kbps is necessary if ISDN is to be used between institution. At this speed the picture-movement is smooth and there is good synchronisation between the lip movements and the speech heard.

The current situation is that if the users in teaching applications are not experienced in the use of IP based videoconferencing then they should weigh up carefully whether they should attempt to use such as system for teaching. The ISDN systems provide a guaranteed bandwidth and therefore picture and speech quality, and they are supported by the large service providers such as BT. Local able companies are beginning to support ISDN. If the service is to be used to other parts of the globe then it is more reliable than the INTERNET because of the guaranteed bandwidth, although it may be difficult to obtain more than 128Kbps in the worst situations. It is not as simple to maintain the picture quality on the INTERNET and it does require people with some understanding of how the system works. No doubt in the near future this will improve, and the INTERNET is more appropriate for the research worker who can tolerate some level of inconvenience from time-to-time. Uni-cast links on the INTERNET are not difficult to establish and maintain, but multi-cast links are dependent on the routers in the network having the software to support multi-casting. This situation is rapidly changing and multi-cast will become commonplace in the near future.

A number of teaching projects have been carried out on both the ISDN and IP networks. In Scotland there are a large number of teaching projects being considered which will use the Scottish MANs. A survey and analysis of these project would be an interesting exercise to try to establish what are the issues involved in network teaching and what a are the lessons to be learned from this experience. The ISDN and MAN projects are following similar approaches in that special rooms have been designed for videoconferencing and these are used for the teaching. The room are not large and much of the teaching concerns small groups. The experience of the INSURRECT project was that this technique was more appropriate for small group teaching and in medicine could be used for the post-graduate training programme where there were large number of students distributed over a large area, but at each individual site there was only a small number of students. This type of teaching requires the facilities for the whole group to see on a large screen the video information being transmitted round the network. The Scottish MAN experiments are using IP for their teaching. Teaching on the INTERNET has been carried out using IP, but this has provided each student with a terminal as in the language training programme RELATE. There is a need fro considerable work to be done is this area, and after the current pilot for teaching with IP on the Mbone, there will be further larger scale projects funded through JISC later this year (1998).

Within the education domain the indications are that desk-top systems, working over the INTERNET can assist researchers and collaborating colleagues to communicate with each other easily and cheaply. The whiteboard is proving to be a very valuable development of this type of communications, where the document under discussion can be seen at each of the participating sites and any changes made at any of the sites can be seen by the other sites. Sometimes people using this type of communications do not use the video channel, using only the audio links and the data link (the document under discussion) - this is the basis of Microsoft's NET-MEETING, which is utilising the T.120 standard.

People should beware if they wish to transmit graphic documents between sites in a videoconferencing meeting or teaching. The ISDN bandwidths are limited and it may be necessary to have one ISDN link for graphics(i.e. data) and a second ISDN link for videoconferencing. ISDN is used extensively for transmitting data, it is not only used in videoconferencing.

4.6.4 The Technical Specification of a Videoconferencing System.

The Technical specifications of a video conferencing ystem conforming with the internationally recognized ITU H.320 standard guarantees that a system will function with any other standardized videoconferencing system:

Room systems,

Rollabout systems, studios

and multipoint videoconferencing bridges, etc.

Computer Requirements

  • PC with Pentium processor or Macintosh with G3 processor (200Mhz plus)
  • H.320 7" PCI format CODEC with built-in ISDN BRI (1 slot)
  • Telephone handset
  • Color camera with integrated microphone
  • Camera connector and power supply
  • Loudspeakers
  • VGA or SVGA screen, or a scan convertor to television screen.

Video Requirements

  • Camera features:
    • PAL color camera
    • Display resolution: 330 lines
    • Pixels: 500 (H) x 582 (L)
    • Min. sensitivity: 20 Lux F2.8
    • 6 mm lens and integrated microphone
    • Swivel mount
    • Adjustable focus
    • Retractable shutter for video secrecy
  • Video input for PAL or, NTSC
  • Video display:
    • 24 bits/pixel
    • 30 pictures/sec. in NTSC mode and 25 pictures/sec. in PAL mode
    • Max. size: 640 x 480
  • Full-motion video compression:
    • H.261 CIF (352 x 288) and QCIF(176 x 144)
Audio Requirements

  • Hands-free or via telephone handset
  • Audio input:
    • microphone or camera microphone
    • telephone handset
  • Audio output:
  • loud speakers or external output
  • telephone handset
  • Software selection and control of audio inputs and outputs
  • Sound coding 3.1KHz in G.711 (56Kbps) and G.728 (16Kbps)
  • Sound coding 7KHz in G.722 (48 or 56Kbps)
  • Lip synchronization with sound
  • Echo cancellation:
    • 256 ms min. in G.711 and G.728
    • 256 ms min in G.722
Still pictures

  • JPEG compression of local picture

Data transmission

  • Dynamic allocation of H.221 frame data channels:
  • LSD 6.4 kbps, 14.4 kbps and 40 kbps
    • HSD 64 kbps
  • X.25 multiplexing with routing of incoming calls

Communication protocols

  • H.221 framing/deframing on two B channels and channel realignment
  • H.230, H.242 protocol management
  • H.231 and H.243 multipoint conferencing

Communication interfaces

  • ISDN (2B+D) daughterboard: 2 x 64Kbps or 2 x 56Kbps
  • MVIP connector:
    • ISDN PCI 2B card
    • ISO ENET card (isochronous Ethernet)
Application software:

The directory, whiteboard and file transfer functions can be used independent of videoconferencing mode:-

  • Videoconference application:
    • Display of local or remote window
    • Picture in picture (PIP)
    • Video secrecy
    • Audio secrecy
    • Clipboard transfer
    • JPEG acquisition and compression of still picture
  • Directory:
    • Automatic audio or video call set-up
    • Multi directory
  • Shared whiteboard:
    • Vector drawing tools
    • Multidocuments
    • Partial screen capture
    • Messaging
    • Remote presentation
  • INTERNET access: Planet PPP

Graphics     Multimedia      Virtual Environments      Visualisation      Contents