The purpose of this paper is to explain in greater detail the H.221 framing structure used in ISDN based H.320 Video Conferences. It explains the framing structures used in 2B calls at 128 Kbps; the differences between 6B and BONDED calls at 384 Kbps and how 56 Kbps Channels communicate with 64 Kbps Restricted Channels.

It is assumed that the reader has a general knowledge of Video Conferencing systems and the standards involved. However, the following technical papers are available to provide more information on these topics:

An ISDN connection has two possible interfaces; a BRI (Basic Rate Interface) or a PRI (Primary Rate Interface). The BRI consists of two circuit-switched B-channels, each of 64 Kbps that are used for data and one D-channel of 16 Kbps that is used for network control. The BRI physically consists of two pairs of twisted wire (transmit & receive) that are terminated by an NTU (Network Termination Unit) in the form of an RJ-45 connector. The PRI is similar to the BRI, but with more channels and extra control bandwidth. In Europe, the PRI consists of thirty 64 Kbps B-channels, that are used for data transmission up to 1920 Kbps and one 64 Kbps D-channel for network control. The PRI is usually terminated by an NTU in the form of two BNC connectors; although some use an RJ-45. In North America, the PRI consists of 23 B-channels of 64 Kbps for data, one D-channel of 64 Kbps for network control and an extra 8 Kbps for Framing.

The H.221 standard defines the transmission frame structure for audovisual teleservices in channels of 64 to 1920 Kbps, as used in ISDN based H.320 Video Conferencing. There are four modes of transmission in H.320, Px64 Kbps, H0 (384 Kbps), H11 (1536 Kbps) and H12 (1920 Kbps); in this paper we will look at the first two modes in terms of 2B (128 Kbps), 6B (384 Kbps) and 384K BONDED (H0 mode) as these are the most common modes used in Video Conferencing.

The signal in an H.221 Frame is structured into 80 octets (8 bits), with an octet being transmitted every 125µs (8KHz). Hence, with each frame consisting of 640 bits that take 10ms to transmit; the total bandwidth transmitted is 64 Kbps. Each bit position within an octet can be regarded as a sub-channel of 8 Kbps. The 8th bit in each octet represents the 8th sub-channel, which is called the Service Channel SC.

In Px64K mode, the first 16 bits of the SC is used for framing and control information. Bits 1-8 are used for the Frame Alignment Signal FAS and bits 9-16 for the Bit Alignment Signal BAS. The initial or only 64 Kbps B-channel is called the I-channel and according to H.221 protocol, Audio can only existing in all or part of the I-channel, with the remaining bandwidth in this and subsequent B-channels being allocated for Video and Data.

In H0 mode, the 384 Kbps channel can be regarded as consisting of six 64 Kbps TimeSlots TS, with the first TimeSlot, TS1 being structured exactly like that for a B-channel and contains FAS and BAS information, whilst the other TS contain no structural information, only data. The TS1 of the H0 channel is also called the I-channel.

With H.320 Video Conferencing systems, each 64 Kbps B-channel can either have its own ISDN number; share the same number as its paired BRI channel or be set so that all BRI channels share the same ISDN number. Usually, a BRI has the same ISDN number assigned to both of its B-channels; this is controlled by the providing Telco. With a 384 Video Conferencing system, the ability to operate in 6B mode or inverse multiplex the six channels together and operate in a 384K BONDED mode is a function of the equipment. Either way, it will have six ISDN numbers that may or may not be the same.

In 6B mode, the 384 Kbps channel can be regarded as consisting of six individual B-channels. When initiating a conference call to a system that is to operate in 6B mode, the six ISDN numbers must all be entered and dialled in the exact sequence that the ISDN lines are connected to the equipment.

FAS and BAS information is present and occupies 1600 Kbps in each B-channel. Hence, in our example, with 56 Kbps allocated for G.711 Audio and 6.4 Kbps for Data, the remaining bandwidth of 5 x (64000-1600) = 312000 bps is available for Video.

In 384K BONDED mode, the 384 Kbps channel can be regarded as consisting of six 64 Kbps TimeSlots, with the first TimeSlot, being structured exactly like that for a B-channel. When initiating a conference call to a system that is to operate in 384K BONDED mode, only the first ISDN number is entered and dialled. The receiving system acknowledges the call on the first channel and replies with its remaining five numbers in the correct sequence. It is therefore crucial that a system is setup to reflect all its ISDN numbers in the correct sequence as this is the only way in which a calling system can determine what to dial. The correct numbers are the local numbers without area code. Only when all six ISDN numbers are known can the dialling system initiate a 384K BONDED call.

FAS and BAS information is only present and occupies 1600 Kbps in the first TimeSlot. Hence, in our example, with 56 Kbps allocated for G.711 Audio and 6.4 Kbps for Data, the remaining bandwidth of 5 x 64000 = 320000 bps is available for Video.

There are advantages and disadvantages to operating in a 384K BONDED versus 6B channel mode; with the obvious advantage being that it can carry an extra 8 Kbps of Video. The conference is also initiated quicker as all six channels are dialled together. With 6B, the channels are dialled sequentially, hence with a long distance call, there is a possibility that the first channel will time-out before the conference is established. The main disadvantage is that all channels must be available for 384K BONDED to work. As FAS and BAS information is only present in the first TimeSlot, there is no recovery mechanism if, for whatever reason, a line is dropped during a conference. If this happens, then the call is terminated.

An important consideration is the effect and use of Gateways and MCUs. Most of these only support BONDED calls at 384 Kbps and do not work in 6B mode. In these situations, you have no option but to use 384K BONDED calls.

To operate in Restricted Channel mode, the Service Channel SC is moved to the seventh sub-channel and the eighth sub-channel set to "ones" in every B-channel or TimeSlot. FAS and BAS are still in bits 1-16, but now in the seventh sub-channel.