MAYAH Communications

Technology Highlights

SIP for compatibility of IP Audio Codecs

Overview

The European Telecommunications Standards Institute (ETSI) was founded 1998, and already one year after there was the DSS1 standard (so called Euro-ISDN) introduced. However it was just the start of the long long process. Audio transmissions over ISDN with professional quality using devices of different manufacturers from different countries was only possible after the long and intensive negotiations between the manufacturers and various standardizing boards, as well as after the number of standards’ and implementations’ improvements.

To avoid the same situation during the establishment of the IP-based connections as the significant communication type for broadcasting industry’s purposes, there was the N/ACIP working group of the EBU founded. The group is already issued the draft recommendations on the audio connections over IP. Based on those drafts MAYAH started the implementations of the newest Audio-over-IP Solutions in its products.

One of the main technology to avoid the mentioned incompatibility difficulties should be the Session Initiation Protocol (SIP). This protocol was standardized by Internet Engineering Taskforce (IETF) and it generally allows the problem-free connections over IP-routes using devices of different manufacturers. Solutions based on SIP are already proved its availability in the Voice-over-IP and telephony industry.

Since there are different algorithms being used in the broadcasting and in the telephony applications, it is important to ensure that the algorithms and their parameters are applied in a standardized form. The N/ACIP working group has already created the general regulations which should be published Autumn 2007.

Manufacturers, who take those regulations in consideration by their developments, allow the smooth communications of their devices to the products of the other companies or to the IP-phones. The codecs of the Audio-via-IP Experts Group members (see also www.audio-via-ip.com) can already communicate to each other using their SIP functionalities.

Insight

An establishing of an IP-based audio connection using SIP usually looks like that:

1.	Codec 1 (e.g. CENTAURI II) suggests linear Audio at 16 bit and sample rate of 48kHz or alternatively MPEG Layer2 with 48kHz sample rate and 384 kbit/s bit rate
2.	Codec 2 (e.g. another SIP-capable codec) is unable to provide linear audio and selects the alterative by responding with MPEG Layer 2, 48 kHz, 384 kbit/s.
3.	Codec 1 ACKnowledges the chosen algorithm.
4.	The media connection is now a using MPEG Layer 2, 48 kHz, 384 kbit/s.

It is quite common to provide more than one alternative encoding so that Codec 2 has several choices. The list of alternatives can be found inside the SDP part of a SIP message and is ordered by priority.

The following shows an excerpt of an example SIP message:

m=audio 5004 RTP/AVP 96 97 9 8 0
a=rtpmap:96 L16/48000/2
a=rtpmap:97 MPA/90000
a=fmtp:97 layer=2; samplerate=48000; bitrate=384000; mode=Stereo
a=rtpmap:9 G722/8000/1
a=rtpmap:8 PCMA/8000/1
a=rtpmap:0 PCMU/8000/1

Here a codec suggests in order of preference:

-  linear audio with 16bit, a sample rate of 48kHz and 2 channels
-  MPEG Layer2 with a sample rate of 48kHz, a bit rate of 384kbit/s and 2 channels
-  G.722 mono
-  G.711 alaw; mono
-  G.711 ulaw; mono

Outlook

During the normal session start with SIP the description of the desired connection is being sent over: two times from the calling codec and one time from the codec which is being called. This is enough for the classical telephony applications, which the SIP originally was aimed for. In such applications there are mostly the algorithms with very few or with no any parameters in use (G.711, G.722, G.723, G.728, G.729a etc).

Within such a short negotiation phase it may be difficult for the devices for high quality media links to agree not only upon the algorithms to use but also the number of channels, sample and bit rates, or if the connection should be a bi- or unidirectional.

To solve such difficulties there are various solutions should be found. For example it may be possible to limit the number of the parameters and algorithms for audio and video codecs to negotiate on. Another possibility could be the extension of the negotiation phase to more than just three steps. So the upcoming recommendations of the N/ACIP working group of the EBU should play the main role in the standardization of the approaches.

see also "Mobility with SIP" »