1.1. Rationale

Almost every contemporary communication system requires one or more signalization channels apart from the channels where multimedia data is exchanged. For example, multimedia terminals must establish the exact moment of the beginning of the transmission to not to lose any data.

The typical scenario of H.323 terminal connection is outlined in Figure 1-1. Two terminals (dotted rectangles) have a special modules (signed as C1..C4), which make up the System Control Unit of the H.323 terminal. The first signalization (called RAS) is performed between the terminal and a so-called Gatekeeper. The RAS link is represented as a curved line (the top of the figure). Two next signalization links are established between presented terminals. They are represented as the two horizontal dotted lines. The upper link is a call signalling link, controlled by the Call Signalling module (C2). It is opened between two terminals to signal that a new connection is requested. If the call signalling procedures succeed (the called terminal agrees to open a new session), a so-called control channel is opened (the lower link). After that the call signalling connection is usually closed (though it can be used later for additional signalling purposes). The two lower links, which are illustrated by multiple block arrows at the figure, represent media transmission, which follows RTP/RTCP protocol.

All the mentioned signalization links have their appropriate functionality. They are defined by different recommendations, for instance RAS and call signalling are defined in ITU-T Rec. H.225.0 [5]. The control channel is defined in ITU-T Rec. H.245 [1], while RTP/RTCP is defined in RFC 1889 [6]. The most complex signalization system is the one related to the control channel (described here). The following functions are performed over the control channel (the list is not exhaustive):

• Exchanging properties of terminals. It is used to inform the remote terminal about supported capabilities. The capabilities are supported audio/video codecs, protocol identifier, and modes a terminal can work in. Once the exchange is successfully finished, a terminal knows what types of audio/video formats the other end can understand.

• Opening and closing logical channels. Once a terminal knows about the other end's capabilities it can choose the format (encoding) of media it wants to transmit. It notifies the other of its choice by opening logical channels.

• Measuring delay in the control channel. Checking if the other endpoint is still alive.

• Establishing and managing maintenance loops.

• Picking the master terminal. A successful completion of some of the activities mentioned above requires designating one of the terminals as "master" and the other as "slave" (there is no democracy in multimedia terminals). Only the master terminal can initiate certain actions.

• Closing the session between the two endpoints.

These tasks cannot be performed without a special protocol. H.323 terminals use such a protocol, the one defined in ITU-T Rec. H.245. The recommendation defines a peer-to-peer protocol, which may be used by a number of different teleconferencing systems (e.g., based on ITU-T Rec. H.310, H.323, H.324 [9]). Not all of the H.245 features are used by any of these systems. The H.323 terminal uses just a subset of all the H.245 features.