hledat:

8   Multimedia Transmissions

8.1   Objectives, Strategy, and Structure of the Project

In the last two years, the Multimedia transmissions project integrated several activities in the area of multimedia data processing and transmission. CESNET is well aware of the importance of supporting advanced applications and this project is thus naturally one of the strategic projects of the research plan.

Our goal has been to create a system for remote collaboration using multimedia applications with a range of audio and video transmission demands. We have aimed to cover different areas of collaboration - from collaboration of individuals to interconnections of specialised centres - by further developing videoconferencing systems, video tools and tools for workspace sharing.

During the last two years project researchers have been focusing on videoconferencing tools suitable for a routine use, systems for implementation of on-demand audio and video transmission and specific applications in the area of asynchronous communication of individuals and groups. The development of network support for multimedia transmission has also been an integral part of this project.

Our effort has been targeted especially on delivering technologies developed by us to potential end users. To this end, we run portals providing basic information that should help end users to select the most appropriate technology and also recommendations and detailed instructions for particular products and possible ways of their utilization. Continuous support of pilot groups and consulting services on multimedia technologies have become an important part of our activities.

The field of multimedia transmissions is evolving quite dynamically and so even after two years of the project we can hardly consider it finished. Despite the multitude of resolved problems we have still many tasks to carry on and new tasks continue to emerge. An important part of our work is an involvement in international activities as well as establishing contacts with similar projects abroad. A close cooperation with researchers working in related areas pursued by CESNET is a commonplace.

The project underwent a large-scale reconstruction at the beginning of 2003. The internal organization was reshaped and we reduced a number of researchers to increase efficiency. The results achieved during the last year suggest this decision was correct. Project researchers published their results in 6 technical reports and 7 journal articles or conference papers. The results were also presented on several important conferences including 3 lectures and 2 posters.

8.2   Collaborative Environment Support

Working on large projects requires coordination and communication between people from the academic centres in the Czech Republic, Europe and often even across the continents. Travelling over long distances is often too expensive and/or time-consuming, while e-mail and telephone communication may not always suffice. A possible solution can then be a videoconferencing service providing real-time video, sound, and shared workspace (editor, whiteboard, ...).

8.2.1   The reflector

Communication of researchers working on large projects often involves multi-part sessions where each participant can simultaneously exchange information with a number of others. One possible implementation of multidirectional transmission is multicast. It can be described as a family of protocols designed to provide management of a communicating group (group creation, login and logout of group members, etc.) and routing of data packets. All the group members receive the data at the same multicast IP address. Data replication and its delivery to all the group members is handled by network nodes supporting the multicast protocols. In particular, the data replication occurs automatically inside the network so that at most one copy of the data is sent over any single line.

This communication scheme brings about increased demands on active components of the computer networks. Current implementations of routing protocols for multicast are often imperfect and cause a high instability of multicast operations, especially in large and heterogeneous networks. It is thus not always possible to deploy a reliable solution based on multicast network services and another more stable system is needed that is also less dependent on the underlying network.

As a consequence of the principles of collaborative sociology and human psychology, communication in a work group typically involves just a limited number of active members. Current high capacity network lines can transfer a reasonable number of data copies without any problem and we can therefore utilize simpler data distribution procedures. One of the possible approaches is data distribution among communicating partners through one central active component (server). Due to the data "reflection" functionality we call such an element a reflector or a mirror. When using a reflector, each of the group members needs just the ubiquitous unicast network connection and no special network services. Compared to multicast, this approach requires larger amounts of data to be transmitted by the network and so the number of participants is necessarily limited. This scalability issue can be seen as the main disadvantage of such an approach.

A reflector architecture must be flexible enough to allow for implementation of all required features and perhaps even capabilities not yet envisaged. We have utilised the following conceptual approaches to achieve our goal:

• active networks with direct and indirect programmable nodes
• overlay networks for implementing specific services over best effort networks
• ability of an end user to create and to manage the service

The result is a UDP reflector fully controlled by the end users. The basic reflector function - data reception and replication for a group of clients - is augmented by a direct management by the end users and a possibility of incorporating other user-provided modules to accomplish specific tasks.

The reflector consists of a number of components responsible for reflector management and data processing. The data processing is structured into modules for data receiving, shared memory, data classification, process scheduling, specific data processing and data sending. Each module for data receiving is connected to a single port. Received data are placed into an incoming queue, classified and checked against certain permission rules (defined by an AAA policy). The session management module maintains a list of client addresses to which the reflector should send the data. At the end of the processing, prepared packets are placed into an outgoing queue and injected into the network by the packet scheduler. The process scheduler takes care about running the modules for data processing while checking the limits of the allocated resources and providing statistical data. The administrative part of the reflector ensures the communication of the user with the reflector through messaging modules, a control module and an AAA module. The communication uses a specific RAP language.

To ensure better scalability, we can interconnect several reflectors using a mesh of tunnels. The reflector management also supports monitoring of relevant events. The actual processing of data passing through the reflector depends on the used modules. There are modules for recording, data transformation, synchronization of streams, or even for combining several streams into one stream to save bandwidth.

Due to the existence of a separate data copy for each client, this environment is also suitable for strong security. This can be achieved by establishing a secure connection for exchanging encryption keys between the reflector and each client. We can also utilize reflectors in a hostile network environment (e.g., for networks obstructed by firewalls). In this case it is possible to transfer data encapsulated in other protocol (either TCP or UDP), which is permitted by the firewall (e.g. HTTP) between two reflectors. The advantage of this solution is that no changes in the configuration of an existing network are necessary.

8.2.2   AGP, PIG, and the Laboratory of Advanced Networking Technologies

At the beginning of 2003, the Laboratory of Advanced Networking Technologies (ANT) was established at the Faculty of Informatics of the Masaryk University (MU) in Brno. The laboratory stemmed from the joint activities of the Faculty of Informatics and Institute of Computer Science at the MU and the CESNET association. It is a research laboratory specialised in advanced networking protocols and applications requiring high-speed networks. The laboratory provides space and facilities for students who work on projects related either to their curriculum or to their bachelor or master theses. The laboratory is also open to doctoral students working on their Ph.D. degree.

The laboratory is currently equipped with a high-end visualization technology, including one 3D and several 2D projection systems and high-end audio facilities. This equipment is hooked together via a programmable RGB and audio signal switch connected also directly to the IP network for remote steering. Equipping the laboratory with these facilities was a part of the effort to build the first Czech Access Grid node. All laboratory premises are covered by both gigabit wired and 802.11b wireless networks. The laboratory provides enough space for new facilities and enough workplaces for students, including two areas with built-in cubicles.

In the laboratory the very first Access Grid node in the Czech Republic has been build. In order to achieve the main purpose of the Access Grid node, i.e., to ensure high-quality communication with similar installations around the world, we need to satisfy a number of conditions. A recommended architecture, including required technical equipment, is available at www.accessgrid.org. Having in mind the fast development in the area of computer technology, multimedia, and computer networks, we decided to modify the recommended architecture so that it would better utilize the available performance and the possibilities of powerful computing servers.

We can see the resulting architecture in the figure. The main modification compared to the standard scheme is a reduction in the number of computers, compensated by a heavy increase of their performance:

• combination of computers for audio and video acquisition and encoding resulted in a single dual-processor computer with the FreeBSD 5.1 operating system instead of Linux,
• combination of the visualisation and control computers into a second dual-processor computer running Windows 2000.

In the opposite direction, the installation of a passive 3D projection (optional for a standard AG node) was an AG scheme extension. Due to the focus on computational chemistry applications and on the research of transmission protocols for synchronous multichannel transmission (multi-channel sound or stereoscopic 3D video), it became one of cornerstones of the laboratory equipment.

All the equipment of the AG node can be remotely controlled. In the next stage we expect to extend the current solution that uses touch panel with our own programmable system. Its web interface will allow to specify and program even more complex scenarios including video, sound, data routing through the network and other elements of the environment (e.g., lighting). The first objective is the development of the software for room control with the following features:

• easy addition and removal of controlled equipment,
• authentication,
• user access control based on a system of roles,
• pre-defined scenarios for particular types of videoconferences.

The software should be integrated with the present AG version 2.0, which is based on the OGSA technology. Another ongoing activity is a research on synchronization protocols for 3D projection and the proposal of system for 3D data storage and sharing.

8.2.3   3D Video Transmission

The research in the area of 3D video transmission in best-effort networks concentrated on the use of DV technology, which appears to be promising thanks to the high number of equipment available at reasonable cost. The advantage of this system is a possibility of implementing a solution without any license fees.

A software solution has been implemented by the DVTS project, which also delivered two RFCs specifying transmission of DV through IP networks (RFC 3189 and RFC 3190). The implementations for the Linux and *BSD platforms comprise tools for reading the DV stream from the IEEE-1394 interface and sending this stream through the IP network. A separately maintained xdvshow program can then be used for viewing received data in X Windows. A Windows 2000/XP version of this display tool also exists and is based on the DirectShow technology.

Recently, the main developer of DVTS left the project and its further development has been stalled. We therefore decided to continue the development of the xdvshow tool which had the worst deficiencies among all DVTS tools. After implementing both direct imaging using the a X11 interface and imaging via the SDL library, we also added support for full-screen mode based on SDL. The new version of xdvshow uses multi-threaded architecture that results both in a more robust data reception from the network and in higher performance enabling the display of two simultaneous DV streams on an ordinary PC without any hardware acceleration. It is thus suitable for software-based display of a stereoscopic video using two independent DV sources.

The 3D scene reading is performed using two DV cameras placed on a stereoscopic tripod head. The data are transferred to the computer via an IEEE-1394 interface and sent over the network with an aggregate data stream of more than 50 Mbps. The display computer receives the data from the network and displays them using two projectors with polarisation filters with orthogonal planes of polarization. The projection uses a special non-depolarising screen, i.e., the polarised light reflected off the screen almost completely preserves its polarisation characteristics. The observer then wears glasses with orthogonal polarising filters that re-create the stereoscopic effect.

8.2.4   Distributed Environment for Video Editing and Encoding

The ever increasing number of requests on video archives in the academic community motivated us to create a pilot editing workplace at the Institute of Computer Science of MU and, specifically, develop an environment for distributed video encoding. The editing workplace is located at the ANT Laboratory premises and is based on technologies from AVID and Adobe. We mainly use the AVID Xpress Pro product, but Adobe Premiere Pro and Adobe AfterEffects are also available.

The output from the editing systems is encoded by a distributed processing environment, typically into the RealMedia format intended for streaming. For this purpose we use the extensive computing and storage capacity of the MetaCenter project. We have developed a prototype of a distributed encoding environment capable of parallel encoding using up to several hundred processors simultaneously. The stream designated for encoding is first stored to a distributed data storage based on the IBP protocol, then automatically split into smaller chunks, which are processed in parallel, and finally joined into the resulting file. Moreover, the available computing power allows complex image transformations (e.g., high-quality de-interlacing or resolution downsampling) to be performed on the fly.

8.2.5   H.323 Videoconferencing Infrastructure

During 2003 we continued to work on the stabilisation and development of the H.323 videoconferencing infrastructure. In particular, we aimed at improving the integration of the existing end stations into the current infrastructure. We distributed several videoconferencing kits Polycom ViewStation and ViaVideo to a number of sites in the Czech Republic. The sites with extensive videoconferencing demands and/or many users are equipped with the stations supporting conferences in the 4CIF resolution (full PAL) and have the functionality of a small MCU.

We cooperate with the manufacturer on solving errors in the current station software and intend to ask them to add new functions in future software versions. As of late 2003, up-to-date versions of firmware providing increased resilience against various network attacks have been installed in most devices.

Currently we are evaluating the quality of the H.264 video codec implementation and the encryption support (AES) in these stations. We expect to deploy these new functions on the major MCUs (Prague and Brno) in early 2004.

An ongoing cooperation with researchers working on the strategic project Voice services in the CESNET2 network resulted in an alternative numbering plan that also incorporates videoconferencing stations. The adopted numbering plan enables stations to access the VideNet videoconferencing network, public telephony network, and CESNET IP telephony network. We are continuously carrying out interoperability tests of videoconferencing stations, IP phones and audio and videoconferencing software applications. We also participate in an inter-project activity that works towards the implementation of a directory services extensions defined by the H.350 standard.

8.2.6   Small multimedia platforms

In 2003 we started the development of our own multimedia platform. Our aim is to create mobile and multi-purpose systems for multimedia transmission focused especially on videoconferences, multimedia acquisition and presentation. Its benefit will be in a simple and uniform control, mobility, low operation costs and reduced noise. We are also developing new acceleration modules, which can make these systems attractive even in the area of special high-quality A/V transmission.

Regarding hardware design we have finished work on two prototypes of this platform for videoconferencing and streaming. Another set of devices is under development supporting the MPEG4@IP transmission using an accelerator card based on programmable hardware (FPGA).

8.2.7   Support of Pilot Groups

We also continue our support to the projects that use videoconferencing facilities (e.g., traditionally DataGrid and IPv6). The experience gained by these groups provides us with important feedback, often indicating weak points in the services provided by our project.

We have created several new types of videoconferencing sets and delivered them to users.

8.3   Special Projects and Activities Support

8.3.1   Cooperation with AVC SH CVUT

Another interesting cooperation that we started in 2003 involves the student group AVC SH (Audiovisual Center Silicon Hill). The goal of this group is to establish a sophisticated semi-professional studio capable of producing and processing high quality digital video - both on-line live streaming and off-line processing of recordings. Their current efforts focus on activities and events organised by the Silicon Hill and the Student Union of CVUT, e.g., OpenWeekend, InstallFest, CryptoFest, and lectures arranged by the SU CVUT. The long-term objective of this group is to implement webcasting of lectures at CVUT. A number of such webcasted lectures have already been organized, most notably the Physics Thursdays, a unique cycle of lectures and seminars arranged by the Department of Physics at FEL CVUT and Programmer Evenings at FEL CVUT as well. All essential information about these activities is available at avc.sh.cvut.cz.

Our contribution to this activity lies mainly in consulting, helping with some scenario preparations, and lending A/V technology.

8.3.2   Recoding and streaming of lectures and other activities at FI MU

Lecture recordings consist of several parts: video, sound and written materials communicated to the students either using a data projector or traditional means like a blackboard. The video part is captured with an ordinary consumer-class tripod-mounted camera (e.g., one of the cameras from the Sony HandyCam series). The camera is connected to the S-video computer interface. We use a PC with an ATI TV Wonder card and the Linux operating system. A virtual driver - so-called video loopback - has been added to the Linux kernel. Inside the computer, the file for streaming and download is created using the program RealProducer 8.51 Basic by Real Networks. Sound is acquired by interconnecting the lecture hall sound system with a sound card in a PC. It is also possible to use sound captured by the camera, though its quality is lower.

Written materials can be captured using the camera. Actually, this is the only option if the teacher writes on the blackboard. Otherwise, if the presented materials are available in digital form, we can insert them directly into the recording and obtain much better image quality. We use the multiplexor software for adding the presentations.

The recordings are available from the web portal video.fi.muni.cz and immediately accessible from any computer in the muni.cz domain. Access from the rest of the Internet requires authentication. One can either directly play the recording by clicking an appropriate link, or download it to the local disk for later playback. The size of an average lecture recording is approximately 300 MB when using single stream with bit rate of 512 kbps. It is thus possible for students to burn the recordings on CDs if needed.

As a test of our abilities in video acquisition and processing, we tried to record on-stage a student performance of Shakespeare's Hamlet. We used several cameras and subsequent video editing. The entire performance was recorded using 3 different DV cameras (Canon XM-2, Canon XM-1, and Sony TRV-30E). This setup led to a subtle complication, since even in the case of prosumer cameras the recording speeds were not identical and we had to correct (re-synchronise) it during editing phase.

We used cassettes with 80-min. tracks in order to avoid changing the cassettes during the performance. The content of the cassettes was transferred through a FireWire interface (IEEE-1394) to a computer disk and stored in the AVI format. The video was saved directly in the DV format without any re-compression (thanks to the fact that same format was used on the cassettes). Minimum requirements for the acquisition computer are thus a sufficiently fast disk and a simple FireWire interface like the OHCI chip by Texas Instruments, which is now often directly integrated on commodity PC boards. As the acquisition software we use Adobe Premiere (version 6.0 or higher). It is also important to have enough free disk capacity - one hour recording corresponds to a file of approximately 14 GB.

The editing has been performed using Adobe Premiere 6.0 as well. Since we are not professional video editors, we spent about 14 hours before becoming reasonably satisfied with the result. This is by no means a stunning efficiency for one hour of video - hopefully the new editing workplace at ICS/FI MU will help us in this respect. We asked the director of the theatre performance to evaluate the edited video. Following his comments, we added front and rear subtitles. The final result was transcoded to the RealMedia format for streaming from the CESNET streaming server in two versions: the first with a maximal quality, intended for powerful computers with a broadband network connection (data stream is around 3 Mbps), and the second one of a lower quality but also less demands on the connection capacity and performance of the receiving computer.

8.3.3   Virtual Participation at APAN Conference - A Telepresentation

Although the ANT Laboratory at the Faculty of Informatics of MU is primarily a research and development lab, the technology is also utilised for supporting regular videoconferences and special events, for example a virtual participation and lecture at APAN conference. This international conference, which is organized by universities and providers of high-speed academic computer networks in East Asia and Pacific areas, was held in Busan, Korea. The videoconferencing tools helped us to participate and even present a lecture without the expensive and time-consuming travelling otherwise needed for attending the conference in person. Of course, part of the time thus gained was amortised in the preparation and technical support of the videoconference.

Conference organisers proposed to use H.323 protocol for video and audio transmission combined with an ordinary presentation in PowerPoint format. As easy as it may seem, we had to cope with unexpected geographical timing issues. Our lecture was included in the session on so-called Logistical Networking but, as it turned out, none of the session lecturers was present in person and all lectures thus were to be presented from remote places. Two of the lecturers were actually in the same time zone as the conference venue (Singapore and Korea) but another was from the USA and yet another (us) from Europe. It was thus impossible to find a time slot comfortable to all participants. Finally, the session took place from 7:00 to 8:30 of local Korean time, corresponding to the interval 1:00-2:30 AM in Brno.

In order to test the videoconferencing setup and network connectivity between us and Korea, we organised two testing sessions in advance. Videoconferences based on the H.323 protocol don't have high bandwidth requirements (typically below 1 Mbps) and we thus supposed the intermediate academic networks would provide enough capacity for ensuring a good transmission quality. However, first test results were an unpleasant surprise for us - network failures and data losses in the network degraded the image and sound quality and rendered it practically unusable. Upon a closer investigation, the connectivity problems were localised on the Korean side and successfully eliminated. A consequent test then resulted in an excellent video and audio quality.

The lecture mentioned above demonstrated that with existing technologies and the capacity of academic computer networks similar activities can be organised with minimum risk. After a careful preparation and with an appropriate technical support (microphones, camera, proper teleconferencing facilities), one can transmit video and audio data through the network essentially to any place connected to the world-wide academic network. With a proper preparation and some training of the lecturer, remote lectures may help to avoid excessive travelling of today's busy professionals.

8.3.4   Education Support at the University of Economics (VSE)

Activities aimed directly at the education support were an important part of the research effort. The creators of multimedia educational applications have to face the problem of a seamless integration of standard video contents with whatever happens on the computer screen. To help with these issues and disseminate knowledge about the solutions we tried, we published a CESNET technical report (11/2003) called "Use of Digital Non-Linear Editing Machinery for Creating Multimedia Education Lessons" (in Czech). In this report we presented the verified and recommended procedures. The proposed technology is based completely on free software thus minimising implementation costs.

Further activities focused on proposing and implementing a technological platform for recording and direct broadcasting of activities associated with education. Already in 2001 it was a technological tool chain for recording and direct broadcasting based on the analog technology. In cooperation with the Department of Philosophy at VSE, we tested the system on live lectures ("Philosophy and science methodology" lectured by doc. Pstružina for doctoral students).

In 2003, based on these experiments and experiences, we have proposed, implemented, and operationally verified a new technology based on digital video. Our results are described in the technical report 16/2003 "Small digital studio". The solution is based on modern digital cameras connected using the IEEE 1394 (FireWire) interface directly to computers and Windows Media Encoder 9 by Microsoft. This technology allows for a simple switching between two cameras with a possibility of inserting video sequences from data files containing e.g., introductory and final subtitles, presentations and optionally other audiovisual information.

The studio technology is mainly software-based, meaning that it uses a minimum of additional hardware components and largely depends on the processor performance. With high-end PCs (Pentium 4, Athlon XP) we can reach the standard PAL resolution in real time. The benefits of the new technology can be seen in the improved quality of the recorded and transmitted image and especially in the simplification of the post-production phase. The technological chain can be applied to recording or on-line broadcasting of lectures by important visitors, seminars, conferences and other similar activities.

8.3.5   Cooperation with Czech Radio

The cooperation with Czech Radio has several levels. This year, we supported a project called "Peregrine Falcons in the Heart of the City", following the tradition of similar projects in the past (Peregrine Falcons in the Heart of the City 2001 and 2002, Millennium Cub, Kristyna Live, etc.). These projects are organized by the Czech Radio and they aim at allowing the general public to look at what is otherwise impossible to observe, and popularise the issue of endangered species protection at the same time. We have constructed a tool chain for audio and video presentation of the peregrine falcon family nesting in Pilsen. Due to problems with the nesting of this particular couple, an alternative locality and content has been selected - Zoo at Chomutov.

Another area of collaboration with Czech Radio is audio streaming. Based on successful tests of a system for permanent on-line broadcasting that we proposed and implemented in 2002, this year we started routine broadcasts of the Czech Radio stations to the Internet in high quality.

The system was initially based on the technology of audio signal transmission in MPEG (MP3) and Ogg compression formats with bit rate 128 kbps, later we added an Ogg stream with a 256 kbps variable bit rate (that means the bit rate is allowed to oscillate around the average value of 256 kbps depending on sound scene complexity). The decision to broadcast in the two formats followed from the fact that while MP3 format is the most widespread in the world and more-or-less automatically expected, the newer Ogg format is considered to be the format of the future. The latter has been developed in a way similar to open source software and is free from the patent and licensing restrictions that hamper free use of the MP3 format. Furthermore, we consider the Ogg format technically superior, able to provide comparable or better quality than MP3 with lower bandwidth requirements. The Ogg format also is supported by the most widespread audio players (e.g. xmms, WinAmp, zinf, qcd, etc.).

Currently we are streaming the following stations: CRo1-Radiožurnál, CRo2-Praha and CRo3-Vltava, all in MP3 at 128 kbps, Ogg at 128 kbps, and Ogg at 256 kbps formats. Since May 2003, we have added the broadcasting of a CVUT student radio - Radio Akropolis. In this case we offered just the capacity of the broadcasting server and the contents are produced by Radio Akropolis itself. They use data streams in the Ogg format with parameters of 44.1 KHz, 16 bits, 2 channels, 128 kbps and variable bit rate.

The operation of the whole system is stable. The dual-processor encoding server works in real time under a constant load of 1.35 while encoding 9 concurrent streams--every encoder instance consumes approximately 13 % of the CPU resources of the server. The streaming server works under a variable load exceeding 0.5 only in case of 100 or more simultaneous clients.

We consider the sound quality to be very good. The highest quality is definitely provided by the Ogg stream with 256 kbps bit rate. With its clear sound scene it is close to the CD quality. In the case of a 128 kbps bit rate, the sound scene is slightly narrower and less balanced. These differences become evident especially with high quality equipment - ordinary computer speakers tend to flatten them. Differences between MP3 and Ogg at 128 kbps are rather minor. In our opinion, though, the Ogg format offers a somewhat higher quality, which can be manifested especially in compositions with a large dynamic range. At the bit rate of 64 kbps, the Ogg format is definitely better than MP3. Judging from the listeners' feedback, the quality of all the streams is perceived as high.

The time delay of the received stream is very small (1-3 seconds) and can be tuned by setting the buffer memory size on the audio client side. While setting a bigger buffer size eliminates the jitter and other irregularities in the received data that are frequent on lines with lower capacity, it obviously increases the delay. For a 256 kbps stream we recommend to increase the buffer size, because the default value corresponds to less then 1 second of play time and this is usually not sufficient.

The streaming servers support both IPv4 and IPv6. IPv4 streams can be selected at
http://radio.cesnet.cz:8000/stream_identification
(e.g., http://radio.cesnet.cz:8000/cro3.ogg)
and their IPv6 equivalents at
http://amp-ipv6.cesnet.cz:8006/stream_identification
(e.g., http://amp-ipv6.cesnet.cz:8006/cro3.ogg).

The following graphs display utilization statistics of the Internet streaming service and document an increasing popularity of this type of service.

As a result of an agreement with Czech Radio, we have access to daily exports of their program schedule in the XML format containing program names and timing data. This information is automatically extracted, processed and added to each stream as a supplementary text data (so-called metadata) describing the program just played.

The issues of selecting technology for such an application are discussed and the final solution described in CESNET technical report number 24/2003.

8.3.6   Piano recital - telepresentation

On November 6th, 2003 we organised a recital of Jakub Litoš, a young pianist and composer, at CESNET premises in Prague. A live stream of this performance was transmitted as a telepresentation over the Internet to audience at the Northern State University (Aberdeen, South Dakota, USA). From the technical point of view, the transmission was implemented as a two-point H.323 videoconference. We used the Polycom ViewStation FX and TANDBERG 880 facilities as the end stations.

The reactions on this event were very positive. Listeners from the Northern State University described the image and especially sound quality as very good. We are thus ready and willing to provide technical support to similar activities again.

8.4   Future Work

Our future plans follow from the CESNET research strategy that also includes support for complex applications utilising the national research network. We want to extend our activities in the following areas:

• completely digital A/V tool chains
• HD capturing, transmission, and displaying
• IPv6 multicast applications
• 10 Gbps end points for A/V applications
• programmable hardware (FPGA, DSP) for A/V
• development of our own A/V platforms
• building new A/V laboratories and AGP nodes
• SIP for A/V and integration with H.323 infrastructure
• methodology for information dissemination, publishing usage scenarios.

previous

contents

next