Session Initiation Protocol – SIP for short – is a modern communication protocol which is currently widely used in audio and video communications. What is special about it and how does it work?

Analog telephony

Not too long ago, telephony was only analog. This required switching though a cable connection from the caller to the person called, which was maintained until the end of the call. Voice data was transmitted in both directions via this connection using analog technology, and signaling for call setup, call tones and ring tones and call termination was carried out with what at that time were relatively sophisticated methods. However, a separate connection was required for each call.

Digitalization

Then digitalization came along, and the invention of ISDN (Integrated Services Digital Network). This offered performance benefits, and at the same time, due to two calls being transmitted over a single line, resulted in increased capacity.

With the advent of the Internet and the gradual changeover of data networks to TCP/IP, demand also arose fairly quickly for conducting telephony communication over already existing data networks. But this called for the invention of new protocols.

• First came the 323 protocol. Although it is similar to ISDN to some extent, and therefore causes some difficulties in the Internet, this protocol is still widely used today. It controls communication from A to Z, i.e. the connection and the payload.
• The so-called SIP protocol (Session Initiation Protocol), which was invented in 1996 with the intention of enabling multimedia applications on the internet, is more flexible. The protocol was later extended for other fields of application, for example for streaming online content or for file transfers. SIP is based on the HTTP protocol, but is not identical.

As the name implies, the SIP protocol is used for call setup. Of course, this also includes connection control and call termination. Other protocols are, however, responsible for the transfer of the payload.

A typical SIP session is fairly simple to understand. Here is an example showing a connection between Sandra and Billy:

Sandra dials Billy's number. Her device sends an INVITE to the interposed PBX (1). The PBX acknowledges the receipt of the INVITE with a Trying, and Sandra's device is informed that the call setup has been started (2). The PBX forwards the request to Billy (3). The PBX also keeps Sandra's device informed by sending a Session Progress (4). For its part, Billy's device now sends a Trying response to the PBX. (5). A Session Progress is then sent from Billy's device (6). This is forwarded to Sandra (7). Now finally the message that Billy's device is ringing (8) is sent. The PBX now informs Sandra that Billy's device is ringing. Sandra then hears the call tone (9). Shortly after, Billy answers the call. His device sends an OK (10). The OK is forwarded to Sandra (11). To ensure that Billy is informed that Sandra has received the OK, Sandra's device sends an Acknowledge message (12). This is sent to Billy by the PBX. The connection is now set up and both of them can speak to each other (13). After a while, Billy ends the call and his device sends a Bye message to the PBX (14). The PBX forwards the Bye message to Sandra (15). This message must be acknowledged with an OK (16). The PBX forwards the OK to Billy, which notifies his device that the call has been terminated correctly (17).

As already mentioned, SIP only includes connection control. The payload is sent via different protocols, which are also established within the SIP protocol during the call setup. In many cases, for voice connections this is SDP (Session Description Protocol), which is responsible among other things for negotiation of voice codecs, and RTP (Real-Time Transport Protocol), with which voice is transmitted.

SIP addresses

So called SIP addresses are also worth mentioning in this regard. Every addressee in a connection must have an SIP address. Typically, they are as follows: sip:user@somewhere.com, or sip:+41798765348@132.78.4.2. As SIP addresses must be used for SIP communication, they can also be found in address directories, for example also in Active Directory.

Incidentally, the session described above is only a very simple connection via a single hop. You can easily imagine what it would look like if the connection is not routed only via a single PBX, but if ten or more network elements are in between.

Other SIP functions

But SIP comprises much more than only the messages shown in the example. For example, additional INVITE messages can be used to forward connections or set up conference calls, or change what is shown on the displays of sets (after call forwarding, for example). The full description of the protocol can be found in a so-called RFC (Request for Comments). All aspects of SIP are explained on 269 pages. The document describes among other things all the messages and responses in detail, what they do and when they should be used.

The SIP Protocol is very powerful yet flexible. (Otherwise, so many pages would not be needed to describe it.) In some cases, the same can be achieved in different ways. For example, what is shown on the display of sets can be changed with INVITE commands. The same can also be achieved, however, with so-called REFER messages, and also with UPDATES.

SIP dialects

This flexibility is a good thing, but often causes problems, because not every provider of SIP devices has fully implemented all facets. If, for example, device A makes changes to the display using REFERs, but device B only knows the UPDATEs, the two are not able to communicate fully, which at best results in minor impairments, but in the worst case scenario completely prevents two-way communication. (In such cases, interposed devices are often required for translation.) For this reason, in this context we often refer to SIP dialects.

In spite of this, SIP is a widely used protocol in Internet telephony, and a number of large telecom providers are about to migrate all remaining analog (and also ISDN) connections to SIP. They would be unlikely to do this if they were not convinced of the advantages of this protocol.

Other topics:

• IP telephony - Opportunities through changeover to UCC
• What you always wanted to know about VoIP
• Requirements for attendant solutions in the digital age