Project Number: Telematic for Research 4007
Project Title: Multimedia Education and Conferencing Collaboration over ATM Networks and Others (MECCANO)
Title of Deliverable:The support provided to validation projects during Year 1 (June 1998 – May 1999).
Deliverable ID: R9.1
Produced by Workpackage: 9
Contractual Date of Delivery: May 31, 1999
Author(s)
Editor: Roy Bennett, University College London
Contributors: Igor Bokun, ACC; John Robinson, CRC; Jörg Ott, TELES; Roy Bennett & Peter Kirstein, UCL.
Abstract:
The MECCANO project, building on the results of the MERCI project,
is further improving the MERCI toolkit to provide a better environment for
multimedia collaboration.
The tools used by the project, whether developed by MECCANO or
from other sources, are proven within the project by the activities of
WP8, Consolidation of Applications. Further validation and important user
feedback is provided by those projects and users using the MECCANO tools
in their work. Some we actively support as part of WP9, Support for
Validation Sites. In this document we report on the work of supporting our
users during the first year of the project.
The projects we have supported have used the tools under the real conditions for remote
teaching, seminars and distributed virtual meetings. We describe the work
of the supported projects and the progress that has been made by them and
by us.
Keyword list:
Multimedia Conferencing, Distance Education, Mbone, User support, Validation.
The Universities of Freiburg and
Mannheim, Germany
The objectives of Work-package 9 are to ensure that the MECCANO software releases meet the needs of the different validation activities using the packages. This includes both fixing problems in the Releases, and ensuring that the contents of the Releases meet the needs of the validation projects. The background to the workpackage is given in the following paragraphs.
By the start of MECCANO, the tools provided by the Multimedia European Research Conferencing Integration (MERCI) project [1] were considered vital to proposals like PROSPECT (ACTS), ICE-CAR (Telematics), NICE (Telematics) and various other european and national projects. Thus in supporting these validation projects we have had toensure that:
The detailed activities have been, and are, driven by the needs, activities and desires of the projects which we support. Although we allocate specific effort to the liaison and tailoring tasks which are described in this deliverable, much effort has also been put into the technical workpackages to support these projects.
We also provide a limited Help Desk facility, via email lists, to provide users of the MECCANO tools with direct help in their use on the various platforms we support and the network environments in which we operate.
In the following pages we summarise the work we have done during the first year in support of the projects. Each of the partners involved in the workpackage has produced a report and the whole is summarised.
The partners in this workpackage are:
We review their work individually in the following sections.
ACC have supported the following user groups which are interested in using video conferencing tools developed or enhanced during the project:
· Medical community of Krakow represented by the John Paul II Hospital;
· A group of scientists from the Faculty of Physics and Nuclear Techniques, University of Mining and Metallurgy, Krakow.
For the purpose of MECCANO WP9, as well as for the future Fifth Framework collaboration, a letter of intention between ACC and John Paul II hospital in Krakow has been signed . John Paul II Hospital (650 beds, 1300 employees) is a specialist medical center with several pulmunological and cardiological wards. The hospital comprises 13 wards, most of them focused on chest medicine. There are also two surgical departments, one for cardiac surgery and the other for thoracic surgery. Two thousand open heart operations are performed in the hospital annually. John Paul II Hospital is one of the most advanced hospitals in the region in terms of the development of information infrastructure. During 1998/1999 the hospital was equipped with the high-end campus network infrastructure based on 622MB/s ATM and switched Fast Ethernet.
ACC provided 2.5 pms of support during the four months to May 1999.
It provides significant support for the fiber link between the hospital and ACC network center. This on-going project, lead by ACC, is scheduled for completion in the autumn of 1999. The 3km high speed link between ACC and the Hospital will allow pan-european conferencing collaboration and telemedicine applications over TEN-155. ACC not only supports high speed Internet connection for the Hospital, but also gives technical support on networking technology, massive data storage and multimedia support. This support includes training on the installation and configuration of MECCANO tools.
As MECCANO tools are oriented towards a wide group of users, ACC has been working on the translation of the VIC, RAT and SDR manuals into Polish. Polish versions of the MECCANO tools documentation allow for easier dissemination of the tools among users with limited English. VIC and RAT manuals are available for free download from the ACC MECCANO web site [2] in the documentation section. The final version of the SDR manual is scheduled for mid June 1999.
Having noticed a substantial interest in Internet video conferencing among the scientific community in Krakow, ACC prepared an introductory seminar on the technology and tools developed under MECCANO. This event took place on 24 April 1999 within the framework of the Open Technical University [3]. The MECCANO project has been introduced, the goal of individual work packages have been described and the most important MECCANO tools have been briefly presented. A PowerPoint presentation from the seminar (in polish) is available for free download from ACC MECCANO web site.
ACC has been also providing limited technical support for users without Internet multicast connectivity. Those support activities included consulting on multicast routing configuration and multimedia hardware and software installation. In cases where full-featured MBone connectivity could not be established, RTP packet reflector (rtptrans) has been used. This particular technique has been successfully utilised in case of the Faculty of Physics and Nuclear Techniques, UMM.
The validation activities have revealed that the current release of the MECCANO tools provides sufficient functionality and stability for the purpose of the collaborative video conferencing over Krakow Metropolitan Area Network. Those of the users evaluating MECCANO tools who had worked with the older versions of the tools noticed a substantial increase in the stability and reliability of the versions released in spring 1999.
An important milestone in MECCANO as far as Poland is concerned was the increase in the bandwidth of the Polish research Internet backbone to 34 Mb/s and its connection to TEN-155 [4] established in March 1999. The improvement of network connectivity in Poland makes it possible to use the MECCANO tools not just on LANs and MANs but also across Poland and the whole of Europe. As many users tend to identify low quality audio and video caused by network congestion with low quality of the tools themselves, ACC's efforts in enhancing network connectivity in Poland, made at the State Committee for Scientific Research level, seems to have had a direct impact on users' opinions on the streaming multimedia tools provided by MECCANO.
An initial contact with scientists representing the Technical University of Gliwice, showing big interest in using MECCANO tools in their projects, has been made. ACC offered technical support in establishing and configuring of MBone connectivity, using MECCANO video conferencing tools, providing multicast traffic monitoring facilities.
ACC's main contribution to MECCANO tools - the Scalable Video Distribution Architecture will be deployed in Krakow MAN. The system will serve as the multimedia repository for the supported user groups, allowing stored multimedia content to be retransmitted on demand.
CRC
has been supporting the MBone Knowledge Networking Project [5] which has 2
interrelated activities:
CRC
has spent a a total of 3.0 pms on providing a regular contribution to
monitoring and trouble shooting the multicast network during the distributed
course sessions. This work has taken place as part of a broader program of
monitoring the national and international multicast support in CA*netII. CRC
has exchanged information on MBone tools with other participants in the
project.
This
CANARIE [6] funded project is nearing completion. There are plans to apply for
support to a follow-on project when the competition is announced. Discussions
about including international partners have been frustrated by the bad
international MBone connectivity from CA*netII.
TELES and UB have closely
cooperated in WP9, particularly in providing hardware and software for
gateways. In various cases support was
given for setting up conference environments using MECCANO tools and/or
instructing users how to use these tools in particular scenarios.
The German CONTRABAND project
dealing with handling of routine work and emergency situations in the harbour
environment has benefitted in its last year from MECCANO tools, infrastructure,
and expertise. In particular, local coordination of MECCANO tools and their
integration with CONTRABAND-specific GUIs and conference controllers is based
upon the Mbus concept developed in MECCANO.
Support was provided to the harbour institutions to set up and configure
a multicast network between the sites involved, based upon their ATM network
infrastructure. Furthermore, some of
the harbour institution personnel were trained to use MECCANO tools and we
gathered feedback with respect to their perceived needs (see below).
Within the Bremen university, a
sequence of distributed lectures was held about media and computer sciences. All these lectures were transmitted via the
Mbone to another Bremen educational institution and to all other interested
observers. Two dedicated rooms in the
two universities were specifically equipped for teleteaching purposes to establish
a virtual classroom. The MECCANO tools
were used for transmission of audio and video between the two primary
sites. The lecturers and the supporting
staff for the lecture series were trained to use the Mbone and to configure
Mbone equipment; also, specialized tool configurations were developed and
appropriate startup files created to simplify the startup procedure for all the
systems. Finally, configuring and
debugging multicasting in the relevant parts of the network were supported.
Within the MECCANO project, partner
sites were supported in installing gateway components. This support includes providing interested
partners with ISDN hardware and drivers for Linux as well as assisting during
the installation process.
The feedback gathered on tool
usability and additional user requirements came primarily out of the CONTRABAND
project and the media and computer science lecture series. In both projects, people felt it essential
to have support for Windows-based platforms that would enable them to use
applications and presentation tools they are familiar with, but were
disappointed by the performance exhibited by the MECCANO tools in the Windows
environment (as opposed to Unix workstations).
In CONTRABAND, the people involved
explained that they would need application sharing as basis for joint
work. Out of the available application
sharing systems for Microsoft Windows platforms, Microsoft NetMeeting seemed
the most workable and best performing solution. As a consequence, we are investigating how to incorporate
Netmeeting into the MECCANO conferencing architecture to be able to provide an
integrated conferencing solution for Windows platforms. For CONTRABAND, it
should also be noted that users had to be urged to try and use the new
technology: the procedure for using the MECCANO tools, even though equipped
with simple GUIs and other simplifying control mechanisms developed in
CONTRABAND, was perceived as too complex while the potential gains in
productivity were not immediately obvious.
In the lecture series, the
lecturers used Microsoft Powerpoint's presentation mode. With all the special
effects provided by Powerpoint, however, using application sharing tools proved
to be insufficient for proper remote presentations: the performance with
animation was not satisfactory and application sharing tools could not convey
all media types used in presentation (e.g. audio clips could not be
shared).
As Powerpoint comes with a built-in
telepresentation mode, we will investigate the possibility of integrating
certain such Windows applications with local control mechanisms such as the
Mbus which is used to communicate between MECCANO applications.
UCL
has put its major effort into the support of the work of the UK project Piloting IP Videoconferencing (PIPVIC2) [7].
This project is a large scale pilot funded by the United Kingdom Education and
Research Network Association (UKERNA) [8] following the success of the earlier
small scale pilot PIPVIC [9].
On
a smaller scale, UCL has worked with the Internet Collaboration Board (ICB) [10]
and the Interworking Public Key Certification Infrastructure for Commerce,
Administration and Research (ICE-CAR) project [11] after its start in January
1999.
The
project started in December 1998 and will be completed in October 1999. Its aim
is to undertake piloting activities to gain a greater understanding of the
issues encountered in running a large scale IP videoconferencing service. It
forms a core part of the UKERNA videoconferencing strategy. The project has
thirteen partners and is coordinated by University College London. The other
Higher Education Institution partner sites are:
·
University
of Birmingham
·
Dundee
University
·
University
of Edinburgh
·
University
of Essex
·
University
of Exeter
·
University
of Glasgow
·
Glasgow
Caledonian University
·
University
of Manchester
·
The
School of Slavonic and East European Studies of the University of London
(SSEES)
·
St.
Andrews University
·
University
of Wales at Aberystwyth
·
University
of Westminster
The
project is active in teaching, research and administration. The teaching
activities provide a number of real courses, including languages, film studies,
sociology, computer networking and business. Research activities are in the
field of particle physics and collaboration in data archiving. Administrative
tasks in which the project is active include external examiners meetings,
project meetings of the several projects, including PIPVIC2 itself.
UCL has spent a total of 2.5 pms on supporting the PIPVIC2 project during this first year. The support provided to the project by MECCANO can be summarised under these four headings:
In each of these areas we were able to provide help and to receive important feedback from users of the tools both on their usability and the general utility of the technology for the purposes of the users.
The actual installation of the tools on PC platforms, which are the most commonly used in the PIPVIC2 project, consists of simply running a self-extracting binary installation file. On workstation platforms the software is provided as a binary file. The source is also provided for local compilation, if necessary.
Most of the problems encountered in this area arose from the variation in the specification of the hardware components in PCs. This variation in the components makes it impossible to ensure that a single version of the software will run correctly on any platform. Typical of these problems were
These problems do still occur occasionally, but, in the main, have only to be solved for a new user or when using a new machine.
We have also experienced some difficulties in the maintenance of version control amongst users. Although we try to maintain backwards compatibility in new versions of the tools this is not always possible. Technical improvements sometimes mandate the use of new versions to make use of new techniques and sometimes there are errors in early versions which can affect later versions in which the errors have been corrected. An example of the latter is the Network Text Editor (NTE) [12] in some versions of which a specific bug is able to corrupt the shared files for all users, even those running a later corrected version. If only one user in a session is using such a corrupting version, the other participants cannot collaborate properly. In an environment of rapidly changing versions driven by user feedback on both bugs and new feature requests, this has proved to need constant attention.
The main requirement which has arisen from working with the PIPVIC2 project has been for changes to the NTE to allow the use of alternative character sets for language teaching. This facility has been implemented for Microsoft platforms (Win95, Win98 and NT4.0) using the Microsoft Multilanguage Support features for these platforms on which the project has standardised. It will only work correctly if all users in the session have the support package installed on their platforms. An additional problem with this feature has been the provision of the correct keyboard mappings. Whilst adhesive labels are available for individual character sets, this is not a scalable solution where there is the need for several such sets as in language teaching. The PIPVIC2 solution for the Russian classes was to have a small keyboard map of the cyrillic characters displayed in a window at the bottom of the monitor.
The ability to synchronise audio and video streams has long been recognised as necessary in an environment where the media are handled by separate tools. In the wide area of the Internet, it has seldom been possible in the past to achieve sufficient bandwidth for video to make lip-synchonistaion necessary. Today, the ability to synchronise is beginning to be an issue once more as more users have access to higher bandwidth networks in Europe. The Robust Audio Tool (RAT) [13] has the facility to use the conference bus feature to communicate with a video tool to allow this synchronisation; at present we do not have a version of the video tool VideoConferencing (VIC) [14] which can make use of this feature. We anticipate that such a feature will be made available in the future.
During the regular weekly project meetings and in the course of other project activities, members of the MECCANO team at UCL have provided analysis of network problems locally and have joined with others at the remote PIPVIC2 sites in collecting statistics. Where problems have been identified MECCANO team members have liaised with UKERNA to solve them. Additionally they have joined with networking specialists at PIPVIC2 partner sites to run diagnostic tests designed to identify structural problems both in the backbone network and in the connections to desktops and conference rooms at local sites.
This has been provided by the presence of a member of the MECCANO project team at UCL whenever an activity involving UCL was taking place. This team mamber has provided technical support and trouble shooting for those taking part. Team members also provide a continuous email help line for the supported MECCANO tools and will respond quickly to requests from users.
Part of the PIPVIC2 project deals with user reaction to the use of the tools and so far there have been a number of analyses produced for specific activities. These reports are available from the PIPVIC2 project web site [7]. In general, user response has been favourable. There has been a positive approach to any problems or deficiencies in the tools which has emphasised a willingness to work with the MECCANO project to provide improved tools with the specific features needed for use in these real applications. Examples of this collaboration were given above in the section Modification of the tools.
We plan to continue the close cooperation which has been so valuable during the past year for the remainder of the PIPVIC2 project.
Edgar Harmsen and Andrew Tu of the Communications Systems Division of the NATO Consultation, Command and Control Agency (NC3A) represented the ICB. The activities consisted of a number of trials over the Internet between NC3A and UCL These trials were held during the period 5-9 October 1998. The trials were documented by the team at NC3A. The report which it produced is included with this deliverable as Appendix I.
The details of the trials may be read in Appendix I. At the UCL end the network support was provided by John Andrews and the tool support by Roy Bennett.
The user response to our collaboration and assistance was extremely positive. This is well illustrated in the summary of the NC3A report which reads, in part:
The trial with UCL improves our understanding of QoS issues concerning multimedia collaboration over the Internet, and promotes international collaboration on related research activities. It is also important to continue collaboration work with UCL which is a both knowledgeable, as well as a reliable and resourceful partner.
A number of the specific points they raised are on our working agenda. For example we have already produced versions of RAT and VIC which operate under IPv6. The group have not requested any further assistance to date.
We have no firm plans to do further work with the ICB, but we remain available to assist them in the future should they ask us.
The objective of the project is to provide all the technology components to support the secure use of the internet for commercial and administrative applications in Europe. The project will improve and deploy the existing security toolsets from the perspective of usability and interoperability.
Since the project started only in January 1999, we have provided little help to the project members in their use of the MECCANO tools. We have liaised closely with them on the improvements which have been made to the Session Direcory (SDR) [15] in its ability to handle secure conferences. An improved version of this tool was delivered as part of the first major delivery of MECCANO tools made in March 1999.
It is too soon, as yet, for any response from participants to their use of the MECCANO tools in the ICE-CAR project.
There is a committment on our part to provide a high level of support to the project members in their use of the tools for conferencing, and we expect to have a high level of informed feed back from them, especially on their reaction to the security aspects of the toolset.
UM and UF provide support for the virtual university project,
VIROR [16]. This project, which started in July 1998, links the universities of
Freiburg, Mannheim, Heidelberg and Karlsruhe to provide higher education
independent of a student's physical location. VIROR encompasses several
departments ranging from computer science to business and language studies.
The most important aspect of VIROR in the context of MECCANO is
that regular lectures and seminars are broadcast as teleteaching events. This
is done as a service for real users - lecturers and students. All VIROR
activities are observed and evaluated by psychologists in order to provide
feedback to tool developers and lecturers alike.
The project uses many of the MECCANO tools including AOFwb, dlb,
wb, vic and sdr.
During the
current semester (April 1999 – September 1999), UM and UF are providing support
for four weekly teleteaching events. In detail these events are:
These events are broadcast over the Mbone. However, a dedicated
ATM backbone (2 Mb/s) is used between project partners to guarantee high
quality and prevent network related unreliability. The toolset varied slightly
between the events and included: AOFwb, dlb, MVoD, sdr, vat, vic and wb.
During all events, UM and UF are responsible for setup and
maintenance of the tools and network connectivity. For the two seminars the
speakers are generally unfamiliar with the toolset and need an introduction on
how to use the tools as well as how to do a telepresentation. This training is
provided in form of practice sessions which are held 2-3 days prior to the real
event. Whenever AOFwb or the dlb are used, the feedback of the participants and
the experience of the technical support staff are used to discover, track and
fix bugs as well as improve the tools with new functionality desired by the
users.
The lectures of Prof. Dr. Effelsberg and the student seminars are
evaluated by psychologists. The results of the evaluation will be made
available as soon as the analysis of the data is complete. Individual comments,
however, show that the acceptance of teleteaching on the student side depends
more on the content and the quality of the presentation of the lecture/seminar
than on technical issues. The lecturers frequently indicated that the process
of getting slides into wb/dlb/AOFwb is still quite complicated, especially when
MS Office (PowerPoint/WinWord) applications are used to prepare slides. Very
noteworthy is the success of the AOFwb recording for last year's lecture on
Computer Networks (Prof. Dr. Effelsberg). The recording is sold to the students
on CD-ROM and is now used by the majority of the students for examination
preparations.
In the following semesters the VIROR teleteaching activity is
scheduled to be increased from semester to semester. The final vision is the
establishment of a virtual university, where students can choose which lectures
they want to attend independent of their physical location. One important
aspect of this location independency is remote examinations. The first such
examination is scheduled to take place at the end of this semester.
During the next semester we plan to incorporate the remaining
tools of the MECCANO toolset into the VIROR teleteaching efforts. Especially
interesting will be the usage of RAT/FreePhone for high quality audio
transmission and MBus for session invitation/setup.
The project has made good progress in the support of projects and users who are actually using the MECCANO tools for real applications, not just for trials. We also continue to respond to the feedback from our users by improving the tools and modifying them to meet new needs.
During the first year our manpower efforts have varied with the pattern of needs of our user community. Some partners, such as CRC and UCL have had an ongoing programme of support at a modest level since the start of the project. Others have applied a more sporadic effort in line with the needs of their particular users. ACC have recently begun a collaboration with a major hospital in Krakow and will continue to develop this relationship during the second year of the project. In the table below we summarise the effort (pms) of the partners.
|
|
Reporting Period |
||||||
|
Partner |
Jun/July 98 |
Aug/Sep |
Oct/Nov |
Dec/Jan 99 |
Feb/Mar |
Apr/May |
TOTAL |
|
C1 UCL |
0.25 |
0.25 |
0.45 |
0.73 |
0.53 |
0.40 |
2.61 |
|
C2 ACC |
|
|
|
|
0.75 |
1.75 |
2.50 |
|
C3 CRC |
0.50 |
0.50 |
0.50 |
0.50 |
0.50 |
0.50 |
3.00 |
|
A6.1 NL |
|
0.20 |
0.10 |
|
0.50 |
|
0.80 |
|
C8
TELES |
0.25 |
|
|
|
0.25 |
0.50 |
1.00 |
|
TOTAL |
1.00 |
0.95 |
1.05 |
1.23 |
2.53 |
3.15 |
9.91 |
Table 1: Effort expended (pms) on Work Package 9.
[1] MERCI Project web site
URL http://www-mice.cs.ucl.ac.uk/multimedia/projects/merci/
[2] ACC MECCANO web site
URL http://www.ics.agh.edu.pl/Meccano/
[3] Open Technical University, Poland
URL
http://galaxy.uci.agh.edu.pl/~amber/Tuo.htm
[4] Trans-European Networking (TEN-155)
URL http://www.dante.org.uk/ten-155.html
[5] MBone Knowledge Networking Project, Canada
URL http://www.acs.ucalgary.ca/~mbone/knprojects.html
[6] Canadian Network for the Advancement of Research, Industry and Education (CANARIE)
URL http://camel.math.ca/Camel/CANARIE/CANARIE.html
[7] Piloting IP Videoconferencing (PIPVIC2)
URL http://www-mice.cs.ucl.ac.uk/multimedia/projects/pipvic2/
[8] United Kingdom Education and Research Network Association (UKERNA)
URL http://www.ukerna.ac.uk/
[9] Piloting IP Videoconferencing (PIPVIC)
URL http://www-mice.cs.ucl.ac.uk/multimedia/projects/pipvic/
[10] Internet Collaboration Board (ICB)
URL http://www.cs.ucl.ac.uk/staff/jon/arpa/icb/
[11] ICE-CAR Project web site
URL http://ice-car.darmstadt.gmd.de/
[12] Network Text Editor (NTE)
URL http://www-mice.cs.ucl.ac.uk/multimedia/software/nte/
[13] Robust Audio Tool (RAT)
URL http://www-mice.cs.ucl.ac.uk/multimedia/projects/rat/
[14] VIdeoConferencing (VIC)
URL http://www-mice.cs.ucl.ac.uk/multimedia/software/vic/
[15] Session Direcory (SDR)
URL http://www-mice.cs.ucl.ac.uk/multimedia/software/sdr/
[16] The virtual university project (VIROR)
URL http://www.viror.de/
A summary of multimedia
collaboration trials between NC3A and University College London, 5-10 October
1998
Andrew Tu
Communications Systems
Division
NATO Consultation, Command
and Control Agency
Abstract:
This document provides
an executive summary of multimedia collaboration trials conducted between the
NATO Consultation, Command and Control Agency (NC3A) and the University College
of London (UCL) over the public Internet.
The test results, a summary of lesson learned and future test plans are
also included.
1. Introduction
As part of our
practical investigation into the operational use of Internet and Internet
technologies and our support for collaboration activities sponsored by the
Internet Collaboration Board (ICB), we participated in a series of multimedia
collaboration trials with UCL over the public Internet.
The focus of the tests was on performance evaluation of
multimedia conferencing services over the IP Multicast Backbone (Mbone)
overlaying the public Internet, using multimedia conferencing tools developed
by UCL. The tests were conducted
between 5-9 October 1998.
2. Test tools
The tests were
conducted using the following UCL-developed Mbone tools:
·
RAT -- A reliable audio
conferencing tool
·
VIC -- A video conferencing
tool
·
WBD -- A whiteboard
collaboration tool
·
NTE -- A text editor
·
SDR -- A session directory
service
·
SHRIMP -- A Mbone tool kit encompassing the above Mbone tools
These
tools were pre-compiled by UCL and we downloaded a copy of the binaries or
executables from a UCL WWW server.
Table 1 summarizes the version of the test tools used in the tests.
|
Mbone tool |
Evaluated Version |
Latest Version |
|
RAT (Robust Audio Tool) |
3.0.28 |
3.0.28 |
|
VIC (Video Conference) |
2.8s5 |
2.8ucl1 |
|
WBD (WhiteBoarD) |
1.0ucl1 |
1.0ucl2 |
|
NTE (Network Text Editor) |
1.5s3 |
1.5.31 |
|
SHRIMP |
V1.0s1 |
|
|
SDR (Session Directory) |
2.4a6s5 |
2.5.6 |
Table1: Evaluated precompiled binary software
version
3. Test configuration
As
our Internet service provider (i.e. UUNET) has not yet resumed the support for
Mbone, a temporary Mbone feed was tunneled from the UCL IP multicast-capable
router to our "pseudo" multicast-capable router which is a Sun
workstation running a special kernel patch (i.e. Mrouted 3.8) supporting
multicast routing. The bandwidth of
this Mbone tunnel was set at the UCL router to a maximum of 128 Kbps.
GSM- and PCM-compliant audio and
H.261-compliant video constituted the audio and video components of a
multimedia conferencing session. The
host machines are PCs running Windows 95 and NT. These PCs were attached to the same LAN segment to which the
Sun-based mrouted router was also attached.
4. Test concept and procedures
We
created a Mbone session that was then advertised to Internet via the SDR
tool. We conducted several Mbone
sessions locally over the unclassified LAN, as well as with UCL over the public
Internet, using the RAT and VIC tool and the SHRIMP tool kit. The tests were conducted during the normal
working hours (i.e. between 8:30 to 17:30) and after the normal working hours
(i.e. before 8:00 and after 18:00). Additionally, we participated in several publicly announced Mbone
sessions. Furthermore, an attempt was
also made to conduct a secure conferencing session with UCL, using the publicly
available PGP-DES facilities.
5. Test results
As
expected the performance or the quality of the multimedia conferencing was
pretty good over the LAN (i.e. with clear audio and about 10 QCIF frames per
second) even during the peak normal working hours. The quality of the conferencing was also found to be acceptable
over the public Internet/Mbone when the traffic intensity over the Internet was
relatively low (i.e. before 8:00 and after 17:30). The quality however started degrading when the traffic load over
the Internet became heavier, noticeable between 9:30 and 15:30. The degradation was exacerbated (i.e. with
frozen video, choppy voice and about 50%-70% packet lost) when the incoming
traffic on the NC3A Internet connection was close to saturated which was caused
by WWW traffic, as indicated by our IP traffic management/monitor manager. This happened about the noon time as people
started browsing the Internet at their lunch breaks. However, this problem only occurred at the NC3A (or the receiving)
end as the quality of our audio and video received by UCL remained pretty good
and nearly the same. This can be
explained as outgoing traffic from NC3A was relatively much lighter than the
incoming traffic over our Internet connection pipe.
Additionally, we experienced some pruning problems, which were then resolved, when a newer version of Mrouted software (release 3.9) was used.
Furthermore,
problems were also encountered during the trials of secure conferencing
facilities using PGP facilities. These
problems (e.g. cannot enter a PGP path at a SDR session setup) are still being
investigated in collaboration with UCL.
6. Lessons learned
·
The UCL Mbone test tools are very useful and are well structured.
·
Bandwidth remains a critical factor for the quality of multimedia
conferencing over the public Internet.
·
Mechanisms capable of providing improved best-effort QoS and
preferential treatment of different traffic streams are required when operating
multimedia collaboration over IP-networks with limited resources.
·
Inclusion of security services such as authentication and encryption
are important for operational use of Internet/Mbone.
7. Near-term test activities
The objective of near-term test activities is to further our investigation into the practical use of Internet technologies for operational use and related QoS issues, as well as to support future ICB collaboration activities. Examples of near-term test activities include:
·
Integration Mbone test tools with COTS multimedia collaboration
packages.
·
Trials of secure conferencing using publicly available PGP-DES facilities.
·
Trials of multimedia conferencing over a VPN.
·
Use of IP bandwidth manager devices for improved QoS through their
admission control capability.
·
Modification of selected UCL Mbone test tools to support IPv6
addresses.
The trial with UCL improves our understanding of QoS issues concerning multimedia collaboration over the Internet, and promotes international collaboration on related research activities. It is also important to continue collaboration work with UCL which is a both knowledgeable, as well as a reliable and resourceful partner. Resources permitting, we should also consider to request our ISP provider to resume Mbone support.