H.320 based videoconferencing in ISDN-, Ethernet- and ATM-networks

 

Jari Isokangas

Tampere University of Technology (TUT)

Digital Media Institute

P.O.Box 553, FIN-33101, Tampere, FINLAND

Email:jari@cs.tut.fi

 

 

 

1 Introduction

In 1990’s the videoconferencing has become an essential part of the distance learning environment. The videoconferencing equipment cost only fractions of what they cost for example 15 or 10 years ago. Systems have also become much smaller, so they do not need a special room for implementation. If you have a PC, a microphone and a videocamera (most of them have a built-in microphone), it is possible to have a videconference on your desktop. In this kind of solution, the coding and decoding of image is done by software which makes it very cheap. The quality of the moving image is not very good but acceptable if you have a good network connection (enough bandwidth).

You can get a much better quality for video if you use a hardware based codec for image compression. In 1990 ITU-T (International Telecommunication Union-Telecommunication) finalized the recommendation H.320 "Narrow-band visual telephone systems and terminal equipment" [Schaphorst 1996]. It defines video codec for audiovisual services at Px64 Kbps (H.261), frame structure for a 64 Kbps to 1920 Kbps channel in audio visual teleservices (H.221), signalling between audiovisual terminals using digital channels up to 2 Mbps (H.242) and frame synchronous control and indication signals for audiovisual systems (H.230). The audio coding for H.320 based videoconferencing system are defined in ITU-T recommendations G.711, G.722 and G.728.

There are other ITU-T recommendations which define same services in networks other than Narrowband-ISDN. H.324 are for low bitrate networks, H.321 and H.310 for Broadband-ISDN/ATM (Asynchronous Transfer Mode) networks and H.323 and H.322 for local area networks (like Ethernet and Token Ring) [Schaphorst 1996].

2 Implementation

In our test we used two different videoconfencing system. BVCS 2.0 videoconferencing system which was manufactured by Bitfield Ltd., Finland [Bitfield 1995], and XtoX integrated videoconferencing system made by Xenex Ltd., Finland. In BVCS the actual codec has two components: H.261 video compression board and audio & framing daughter board. The system also needs an video overlay card to be used with regular video display card or Bitfield’s own Digital Video Prosessor (DVP). In DVP both of these functions are integrated in one card. XtoX system is based on a new Mediamaster videocodec, which can be connected directly to the video card and does not need a separate overlay card. At the moment XtoX can be used only in ISDN environment. For voice we used videocameras with built in microphones, separate microphones, active speakers with built-in amplifiers, and normal HIFI amplifiers with passive speakers. For video image we used normal VHS-C and S-VHS-C videocameras and a data projector.

We installed the BVCS videoconferencing system to WIN95 PC’s which had different network interface cards (NIC). For ISDN-based network we used SCII Expresso card which offers Basic Rate ISDN (BRI) connection with BVCS. In pure local area network we used regular Ethernet NICs with twisted pair and coaxial interfaces, and in ATM-network 25 Mbps twisted pair NICs and 155 Mbps twisted pair and fiber NICs.

Only one type of connection can be used at the time. If you want to change for example ISDN connection to LAN or ATM connection you have to load different Connection and Transfer Kerner (CTK). In oreder to do this you have to change the CTK entry in BVCS.INI and restart BVCS.[Bitfield 1995a] In ATM based videoconferencing there was a TCP/IP protocol stack over ATM layer, so it was not native ATM applications even both participants were directly connected to ATM switches.

ISDN connection

In ISDN network, the maximum bitrate we were able to use with our configuration was 128 Kbps (2 x 64Kbps). The quality of video image was much better with XtoX system than with the older BVCS system. We had ISDN-based videoconferences to several places in Finland and abroad with both equipment. There were some compliency problems with systems that have different manufacturers. In most cases video image and voice was satisfactory but other features like echo cancelling, did not work in both ends. There were more problems with the new XtoX than with the older BVCS system. We found that the quality of image and voice is good enough for personal communication. You can also send pictures and written text through video, even though this is not recommended if you do not have another fixed camera for that use only.

LAN and ATM connections

We tested videoconferencing in pure local area networks, Classical IP networks [Laubach 1994], LAN emulated ATM networks [Finn and Mason 1996], and tried also different combinations of these. We also tested videoconferencing through public Internet. In traditional local area networks and local ATM networks we used transmission rates between 128 Kbps and 2 Mbps, and in Internet connections the maximum speed we set was 1 Mbps. But in connection through public Internet you can not reserve the bandwidth, and the actual bitrates were greatly varied. We tested the videoconferencing over ATM in our campus network as LAN conferencing. Through Internet we had videoconferences in Tampere area and to Jyväskylä.

The quality of picture was better in most of the cases compared to ISDN connection, and the voice quality was equally good as in ISDN connection. The voice is more sensitive for delay and variation of delay than video and in some cases when there was much traffic in public Internet, the quality of voice was very bad and it was not possible to understand what other participants were saying. Using the maximum bitrate in ATM network the video quality was excellent. The quality of video did not improve in accordance with the increase of bandwidth. The improvement was much bigger between 128 Kbps and 400 Kbps than between 400 kbps and 2 Mbps. Comparing to ISDN there was a big difference when sending picture and text through video. You could read much smaller text with ATM than with ISDN connection. In traditional LAN environment there is a big difference between using shared or dedicated connections. With dedicated connections you could have the same 2 Mbps bandwidth for videoconferencing as in ATM network, but you can not be sure that you have it through the entiry conference. In mixed environment the connection between participants is as bad or good as is its weakest link.

3 Conclusions

The quality of voice and image is rather good with modern videoconferencing system. As was expected, the best possible quality is achived, when using ATM from end to end in LAN or WAN environment. On the other hand you could have same quality in your own LAN environment when you are using Ethernet switches and dedicated connections for all clients. At the moment, the ISDN based solution is most reasonable for the WAN videoconferencing. The quality of video with one ISDN line is satisfactory for personal communication, but if you could have 3 ISDN lines (384 Kbps) the quality of video would be fairly good.

4 References

[Bitfield 1995]Bitfield Video Communication System: Version 2.2.

[Finn and Mason 1996] Finn N., & Mason T. (1996). ATM LAN Emulation. IEEE Communications Magazine, June 1996, 96–100.

[Laubach 1994] Laubach M. (1994). Classical IP and ARP over ATM. IETF RFC 1577.

[Schaphorst 1996] Schaphorst, R. (1996). Videoconferencing and videotelephony: technology and standards.Norwood, MA: Artech House Inc.