The different subscribers require unique addresses to be able to establish communication connections. The structure of these addresses is defined in SIP. An SIP address consists of a URL and domain name, similar to an e-mail address. An example for this type of address is “firstname.lastname@example.org". The first part of the address is the user name or often times the telephone number. The domain specifies the respective Session Initiation Protocol network.
In Voice-over-IP telephony, Session Initiation Protocol merely controls establishing and closing the connection as well as the connection modalities. The actual speech information is transmitted directly between the subscribers using a different protocol. VoIP typically uses the Realtime Transport Protocol (RTP, RFC 3550) for this purpose. SIP uses SDP (Session Description Protocol) to define details of the actual media transmission between the subscribers via RTP. The connection is established with an INVITE message, acknowledged with an OK message. The subscribers must first register their SIP addresses with a registrar server if they do not recognise each other.
Key elements of the SIP system architecture are the User Agent, the Registrar Server and the Proxy Server. The User Agent is a terminal such as a telephone, computer or mobile attempting communicate via Voice-over-IP and which has an SIP address. If they recognise each other, they can communicate directly. The Registrar Server controls locating the subscriber. For this purpose the User Agents regularly report to the registrar using their SIP and IP addresses. Using this information, the server is able to address connection requests. The Proxy Server lastly can be used as a client or server and send requests on behalf of a client. Other important elements in the SIP architecture are Redirect Server, Session Border Controller and Gateways.