Pekka Isto 13.11.1998 1(8) Internet Desktop Video Conferencing ABSTRACT: This is report outlines possible use of Internet desktop videoconferencing software in a distributed engineering project and presents the experiences of the GECOS team at Helsinki University of Technology. DOCUMENT INFORMATION: Project: GECOS Document: Internet Desktop Video Conferencing File: D:\users\evp\GECOS\WIP\conf1111a.doc Deliverable: Publicity: GECOS Version: 0.9 Status: Draft Date: 12.11.1998 Author(s): Pekka Isto Reviewed by, date: Checked by, date: Approved by, date:
Pekka Isto 13.11.1998 2(8) TABLE OF CONTENTSINTERNET DESKTOP VIDEO CONFERENCING 1....INTRODUCTION 3 2....CAMERAS 5 3....NETMEETING 5 4....CU-SEEME 6 5....IVISIT 6 6.... NETSCAPE CONFERENCE 6 7.... EXPERIENCES AND CONCLUSIONS 7 8....REFERENCES 8
Pekka Isto 13.11.1998 3(8) 1. INTRODUCTION Internet in it s early forms was principally a text and file transfer medium. Increased bandwidth and popularity have enabled a set of new forms of communication over the Internet. Among the most interesting is real-time video conferencing. While video conferencing has been long known as a special service over ISDN phone lines, the Internet variation of video conferencing has two important advantages. Internet conferencing systems are based on personal computers and thus have the potential to become readily available tool on the personal desktop. The cost of data communication over the Internet is typically much lower than over dedicated (phone) lines. Together these features may result to affordable videophone on every desktop. Effective communication between the engineering team members is a crucial issue for the success of the team. Formal and informal meetings are usually the main method to distribute deep or background knowledge and build a shared understanding about the mission of the team. For a geographically distributed team the arrangement of meetings and travel may rapidly become a burden, often to the extent of making it unproductive to establish the distributed or virtual team in the first place. Teleconferencing in the forms of audio, video and data conferencing may provide a means to reduce the need for face-to-face meeting. Teleconferencing enables planned and spontaneous communication between members in a distributed team and enables additional live interaction below the reason-to-travel threshold. Since engineering involves complex data sets, which are difficult to describe verbally, engineering sessions are difficult to carry out with audio only telephone conferencing. Graphical data and in some cases video is essential for complete understanding of the design issues. Predistributed material may be used for structured design reviews, but for rich two-way communication live sharing of graphical and other data between the participants is required. Conferencing tools should be deployed as widely as possible to let potentially every member in the enterprise to enter the conference. Input from sales, marketing, manufacturing and other non-engineering departments can be valuable in the design phase but usually is affecting the (distributed) engineering only indirectly due to travel costs. Inexpensive personal computer based conferencing systems show a great promise here. Teleconferencing also has the advantage of being easy for the passive members in the meeting. The behaviour of the listeners do not disturb as easily the flow of the discussion among the principal active members of the session and the listeners do not feel pressure to participate for the sake of making a conversation. The available conferencing tools provide functionality for video and audio transmission, text chat, shared whiteboarding, shared clipboard, file transfers, application sharing and collaborative WWW browsing. While video and audio transmission is used mainly to communicate speech and facial expressions, they may be used to communicate other events at the participants sites. Text chat is
Pekka Isto 13.11.1998 4(8) simply exchange of short text messages on a shared text area. The messages are typically prefixed with the writer s name or other identification tag to keep track of the conversation. Many systems support also private messages directly from one participant to the other. Shared whiteboarding is an electronic version of the plain old whiteboard. User may collaboratively write, draw and post images on the whiteboard and the system keeps the participants whiteboards synchronised to the changes. Shared clipboard allows users to pass information between shared and local applications. Users may distribute files during the conference with the file transfer support. Application sharing lets conference participants to work with the same software application and document. Collaborative WWW browsing is a specific form of application sharing. A single software product offers only a subset of the above functionality. Furthermore, the products from different vendors are often incompatible even when they support the same or similar functionality. Currently, there exists a set of International Telecommunication Union (ITU) standards for telecommunications tools. The relevant standards for Internet conferencing are T.120 series for application sharing and H.323 for multimedia applications (audio and video conferencing) over networks that do not provide guaranteed bandwidth. Some products, however, implement the standards in non-compatible way or support some proprietary protocol. Systems conforming to H.323 standard require a Multipoint Control Unit (MCU) to support multiparty conferences. This server takes the responsibility of broadcasting the audio and video streams between the participants in the conference. Each participant uses the client or terminal software such as Net- Meeting to connect to a conference running at the server. Proprietary solutions for multiparty conferences are products in the CU-SeeMe family and ivisit. CU- SeeMe multiparty conferences are run via a type of server called Reflector. Unlike most other products, ivisit lets users run multiparty conferences with just the client software, no special server is needed. For real-time full motion video, a minimum bandwidth of 128 Kbps (2 ISDN B channels) is recommended for reasonable image quality [1]. Six B channels or 384 Kbps is sufficient for VHS-like image quality at the resolution of 352*288 pixels [2]. The video and audio quality and bandwidth requirements depend on the method used to transmit the signal over the network connection. These methods are usually referred to as codecs. Most products provide a set of codecs for users to select for optimal balance between quality and network load. In many cases the quality of the video image is non-essential and can be sacrificed without much effect on the quality of the conferencing experience itself. The audio codecs support bitrates from 5 Kbps to 64 Kbps. The conferencing software requires usually a Pentium class computer with multimedia support. The GECOS team at Helsinki University of Helsinki has performed a set of experiments with freely available and inexpensive Internet conferencing tools. This short report is set to present these products and experiences. The tests were intended to evaluate the usability of inexpensive PC peripheral cameras and various pieces of software for communicating over networks of various complexi-
Pekka Isto 13.11.1998 5(8) 2. CAMERAS 3. NETMEETING ties. Tests were done in two network environments. The first was a single switch connecting the participants with 100 Mbps Ethernet connections. The second network connection was more complex involving several switches and fiber-links. There are several inexpensive, low-resolution digital video cameras marketed as PC peripherals for video conferencing. The GECOS team purchased two for evaluating the usability of the video conferencing systems. The selected cameras are KODAK DVC 323 and Alaris QuickVideo DVC1. Both are intended for PC compatible hardware running a version of Microsoft Windows. KODAK DVC 323 is a colour video camera with USB interface and drivers for WINDOWS 95 OSR 2.0 and WINDOWS 98. It supports the following set of video resolutions: 160*120, 320*240, 160*120, 320*240, 176* 144 (QCIF) and 352*288 (CIF). The frame rate of the video stream depends on the selected resolution. Rates upto 30 frames per second (fps) are possible. The camera comes with TWAIN-compliant drivers. KODAK DVC 323 requires at least a Pentium based computer with 16 MB of memory. Alaris QuickVideo DVC1 is a parallel port colour video camera for Windows 95, Windows 98 and Windows NT 4.0 computers. The minimum hardware requirements for DVC1 are a 100 MHz Pentium with 16 MB of memory and bidirectional parallel port (ECP parallel port required for optimal performance). The available video resolutions are 160*120, 240*180 and 320*240 with frame rates from 15 to 30 fps. DVC1 is also provided with TWAIN compliant drivers. Microsoft NetMeeting is perhaps the best-known product for Internet video conferencing. Microsoft provides the client software for free. They claim that over 25 million copies of the various versions of the software have been downloaded [1]. NetMeeting provides video conferencing, Internet phone, text chat, multipage shared whiteboard, shared clipboard, application sharing (T.120 standard) and file transfers. NetMeeting supports multiparty meetings, but users may receive only one video and audio stream at time and must manually switch between the participants. A H.323 standard compatible conference server such as White Pine MeetingPoint is required for a genuine multiparty conference. The application sharing is implemented by redirecting mouse and keyboard commands from the controlling participant to the host of the shared application and sending bitmap images of the application to the participants. This means that only the host of the application sharing session has to have the application installed. On the other hand, the sending of the bitmap images may make the interaction sluggish. NetMeeting is available only for Windows 95 and Windows NT 4.0.
Pekka Isto 13.11.1998 6(8) 4. CU-SEEME 5. IVISIT 6. NETSCAPE CONFERENCE The version of CU-SeeMe released from the Cornell University is one the first wide spread video conferencing products on the Internet. The Cornell CU- SeeMe is still available for free downloading. It supports grey-scale video over a proprietary protocol. For multiparty conferencing a special server called Reflector is needed. The participants connect to an agreed upon Reflector and join a conference. CU-SeeMe can be used as a multicasting client, but other newer products are preferred now for multicasting video. Free Reflector software is available from Cornell University and Brian Godette for several platforms. White Pine enhanced CU-SeeMe is a commercial version of CU-SeeMe. It supports colour version of the proprietary CU-SeeMe video conferencing protocol. Furthermore, it supports text chat, and shared whiteboarding is supported as a separate program. The whiteboard is arranged to pages in a workbook allowing a more structured session. Enhanced CU-SeeMe can participate in H.323 conferences via the White Pine MeetingPoint conference server. Enhanced CU- SeeMe is supported on Windows 95, Windows 98, Windows NT 4.0 and Macintosh System 7.6 (or later). The most interesting feature of ivisit is that it supports multiparty conferences without any centrally managed server site. However, the functionality provided by ivisit is very basic. It offers just colour video and audio conferencing and text chat. The software is downloadable for free, but it is programmed to stop functioning after a certain date, so that the user has to download a new version periodically. This is suggests that ivisit may become non-free at some time in the future. Supported platforms are Macintosh System 7.5 (or later), Windows 95, and Windows NT Netscape Conference is a part of the Communicator 4.0 suite of applications. It is not really a video conferencing product, since it does not support video transmission. Supported features include H.323 compatible audio conferencing, text chat, shared whiteboard and collaborative WWW browsing. The shared whiteboard supports only one-page documents. As a H.323 compatible product Netscape Conference should be compatible with H.323 conforming conference servers, but without video stream. Text chat and whiteboarding sessions are possible only between users of Netscape Conference. Audio conferencing is interoperable with NetMeeting 2.1 on LANs, but there are not compatible low bitrate audio codecs for interoperation on modem connections. The advantage of Netscape Conference is that it is available for wide variety of platforms including several flavours of Unix.
Pekka Isto 13.11.1998 7(8) 7. EXPERIENCES AND CONCLUSIONS Experiences with NetMeeting versions 2.0 and 2.1 were successful over simple network architectures. Point-to-point calls were successful and the quality of the audio and video were sufficient for performing some simple simulated negotiations. The application sharing works as advertised but is somewhat slow even with the fast Ethernet connections. Over a more complex network architecture, the video and audio qualities decrease noticeably, probably due to network latencies. The frame rates of the video stream were quite constantly below 10 fps. While the video quality is not a very important issue with technical sessions, the breaks and other disturbances in the audio stream are very annoying. Unfortunately, the author did not succeed to configure NetMeeting to use the Alaris camera on his computer. However, the installation procedure was successful with a similar but not identical computer set-up. The reason for this disparity remains to be identified. The switching between meeting participant could not be fully tested, since there were also problems with the audio set-up of the author s computer at time of the tests. It appears like the configuration of the software is not a trivial matter and involves some effort, if the default settings are not satisfactory. The Cornell and Enhanced versions of CU-SeeMe were tested for point-topoint calls and conferences with local and remote reflectors. Again, the local connections with minimal network latencies were satisfactory with comparably video and audio quality to NetMeeting, but longer connections suffer from variation in the network load. For some unidentified reason the Cornell CU- SeeMe did not connect to the Brian Godette s Enhanced Reflector unless there were already at least one connection to the conference from the White Pine Enhanced CU-SeeMe client. This was probably a configuration problem. Sessions with ivisit were notable because of the high quality of the video stream. Over the local switch it was possible to get almost as good video stream as the camera was capable of providing. Longer connections were carried out as video and text chat sessions, so the quality of the audio stream remains unevaluated. However, as far as the author understand the issue, there are no fundamental technological differences between the various audio codecs. Therefore the same problems with audio quality are probably evident with ivisit too. Netscape Conference was used only to test the interoperability of NetMeeting. An audio call between Netscape Conference and NetMeeting over a LAN connection was successful as expected. The overall impression from the tests was that the most important feature of the evaluated software was the data conferencing. It is the author s understanding that live video stream is seldom essential for engineering meeting, but the quality of the audio is very important. These software products essentially trade the audio quality for the slow motion, low-resolution video stream. A possible way to take advantage of the best features of these products is to combine the data conferencing features with conventional telephone conferencing. That would
Pekka Isto 13.11.1998 8(8) 8. REFERENCES allow shared whiteboarding and application sharing to communicate issues that cannot be easily described in speech only and simultaneously enjoy the good audio quality provided by modern telephone equipment. Most of the products described in this report are available only for Windows and Macintosh platforms. This is a limitation, since many of the engineering tasks are still carried out with Unix based workstations. There are similar products available for Unix platform [3], but these are often expensive systems and were therefore not evaluated here. Security is an issue when the conference participants are discussing confidential and in some cases valuable strategic issues. None the evaluated products support any kind of data encryption. The entry to a conference may be protected with passwords, but the data streams can be captured and decoded by persons having access to a computer along the route of the data streams. Confidential conferences should be held within trusted Intranets or protected with Virtual Private Network and other tunnelling methods. The data protection by encryption surely adds more delay to the network connection and may further decrease the quality of the service. [1] Amitava Dutta-Roy, Virtual Meetings with Desktop Conferencing, IEEE Spectrum, July 1998, 47-56. [2] Videoconferencing FAQ, http://www.bitscout.com/faqtoc.htm [3] List of video conferencing products, http://www3.ncsu.edu/dox/video/products.html KODAK DVC 323: http://www.kodak.com/us/en/digital/geninfo/dvc323.shtml Alaris QuickVideo DVC1: http://www.alaris.com/hardware/harddvc1.htm NetMeeting: http://www.microsoft.com/netmeeting/ CU-SeeMe: http://cu-seeme.cornell.edu/ Enhanced CU-SeeMe: http://www.wpine.com/, CU-SeeMe reflector: http://actuva-www.larc.nasa.gov/vc-help/curef.htm Enhanced CU-SeeMe reflector: http://www.dimensional.com/~bgodette/ ivisit: http://www.ivisit.com/ Netscape Conference: http://home.netscape.com/communicator/v4.0/datasheet/standard.html#conferen ce