VII. VIDEOCONFERENCING DEMONSTRATION
A. INTRODUCTION
B. OVERVIEW
C. DEMONSTRATION
D. RECORDING A BROADCAST
E. PLAYING BACK AN MBone RECORDED
SESSION
F. EVALUATION OF RESULTS
G. SUMMARY
A. INTRODUCTION
This chapter provides a proof of concept that demonstrates how current videoconferencing software can be used to deliver synchronous or asynchronous material for distance learning over an IP based network via multicast. The demonstration is follow-on work accomplished in Internetworking: Economical Storage of and Retrieval of Digital Audio and Video for Distance Learning [Tiddy, 95] and Internetworking: Worldwide Multicasting of the Hamming Lectures for Distance Learning [Emswiler, 95].
B. OVERVIEW
Several free software tools were considered, and the one selected was the MBone VCR on Demand (MVoD), developed by Wieland Holfelder at the University of Mannheim, Germany. The MVoD is a free, experimental software solution for the interactive remote recording and playback of multicast videoconferences. The MVoD Service offers a graphical user interface (GUI) environment where the user can interactively record audio/video conferences on a remote server, controlling the recording session with a local client application. Later, that same user or other users can play the session back on demand, via multicast or unicast.
Through the use of this tool, the goals of this experiment was to demonstrate:
To begin the testing, the MVoD Service software was downloaded from the site http://www.informatik.uni-mannheim.de/informatik/pi4/projects/MVoD. Version 0.9a7 of the software was installed on a Silicon Graphic Indy, running IRIX 6.2 OS, 128 MB of RAM, running a MIPS R1000 processor. The MBone tools sdr, vic and vat were already installed. The MVoD architecture consists of three components:
A number of internal protocols have been developed to provide communication between the various MvoD software components. They include the:
An interface has also been implemented with the Session Announcement Protocol [Perkins, 97], which is used by the MBone tool sdr, in order for the MVoD server to learn about ongoing MBone sessions. Figure 7-1 is the MVoD architecture with its various protocols, which are used in conjunction with MBone tools. Detailed explanations of the various protocols can be found at the web site.
The testing was accomplished using two SGI workstations (Indy and Octane models) on the NPS LAN. The test lectures for the multicast transmission, which had been developed from the thesis "Internetworking Worldwide Multicast of the Hamming Lectures for Distance Learning" (Emswiler, 95), were input to an SGI Indy workstation (blacknoise) from the line output of a VCR . The MVoD Service was running on an adjacent workstation (electric). The MVoD Service, and the MBone’s sdr, vat and vic were the software used for the experiment. The MBone tools are also free and can be downloaded form many ftp sites that provide MBone tools. The MBone tools used have already been proven effective, therefore the focus of the Chapter will be on the effectiveness of the MVoD recording and playback processes.
The first step was to set up the workstation expected to multicast the lecture over the MBone by providing video and audio line connections from the VCR. The sdr tool on blacknoise was used to create a new MBone session. The video and audio source of the MBone transmission was provided by a VCR that played back the Hamming Lecture Series. Once the VCR was connected and the session created, the lecture was multicast over the MBone using vic and vat (RTPv2). For the multicast, default bandwidth settings for vic H.261 (128Kbps) and vat PCM audio (64Kbps) were used. The time-to-live (ttl) was set to 15, in order to keep the transmission restricted to the campus LAN.
On the workstation "electric," the MVoD Service was running. The MVoD client GUI was used to control the MVoD server and RTP data pump. The May 26th 1995 lecture was recorded by MvoD using a 128kbps (maximum) vic video stream, which lasted for 37 minutes. After the session was recorded, the file size of the recording was noted. Based upon the five files that the MVoD server creates for each recorded session, the total file size for the transmission was approximately 46 MBs. Therefore the recording averaged 1.24MB per minute of data stored. If a standard 50-minute lecture were held, the storage requirements would be approximately 62 MBs. An expected file size is thus approximately 75MB per hour. (This size fits conveniently inside of a 100MB zip disk).
E. PLAYING BACK AN MBone RECORDED SESSION
The next step in evaluating te MVoD Service was to play back (multicast) the recorded session while simultaneously transmitting it over the MBone. Using the MVoD client GUI on electric, a list of sessions previously recorded (which was only one, in our case) by the server was displayed. Once the session was selected, the GUI also provided the option of playing back either audio, video or both mediums. Both audio and video were selected. When the play button was clicked, the session was multicast over the MBone and vic and vat were automatically launched locally in order for the person playing back the session to observe it. The transmission used the same bandwidth settings that were used during the original session and can not be changed.
The rebroadcast (play back) of the session was observed using vic and vat tools on "electric" and "blacknoise." From the observation, there was no discernable difference between the recorded session and the original. There was no packet loss due to the fact that there was no congestion on the LAN containing the multicasting and receiving workstations.
Currently MvoD only runs on UNIX systems. For a user having little experience with UNIX command lines and environment variables, the MVoD tool is not easy to install. Therefore it is recommended that only System Administrators or experienced UNIX users install the software. During the initial installation, there were problems with killing processes. For instance, some processes could not access sockets even after prior process at the socket had been killed. After becoming more proficient with the tool, and properly shutting it down, this was no longer an issue. Further development of MvoD may make installation simpler, and will likely provide a Windows version as well.
One result important to note that having too many applications running on the workstation slowed down the CPU cycle time, effectively slowing down the compression rate of the transmission. In all of the playbacks (multicasts), the default transmission rates on the audio and video provided a clear reproduction of the original audio/video session. No experiments were conducted using the using the MBone wb tool. Whiteboard recording is not likely to occur soon due to the distributed asynchronous nature of events.
The results of the audio and video testing are satisfactory and demonstrate the successful recording and payback (multicast) of a distance learning lecture using the MVoD Service.
The results of this experiment proved that the technology exists for software tools available to receive, archive, and retransmit distance learning lectures. Once set up properly, the software provides a simple GUI that is easy to use, and not only provides playback on demand but also recording on demand. Being able to record content for future use enables users to build a local library of distance learning content.
The MVoD tool, or a similar tool, can be used to remotely record an instructor’s lecture. MvoD could be set up to perform as described in Chapter V. A student can use the MBone tools to connect to the session during the live broadcast, or use the MVoD client GUI to receive a prerecorded session at a more convenient time. If bandwidth over the network segment is restricted, which may often be the case, users can ftp the session from the cache server for local playback.
Table of Contents | Previous Chapter | Next Chapter | Main Thesis Page | Seanet Page