SDP : protocole de description de session

SDP stands for Session Description Protocol. It's a text-based format used in WebRTC to describe the capabilities and parameters of a media session.

Think of it as a blueprint for what kind of audio, video, and data a peer wants to send or receive, and how it can be reached.

Before two WebRTC peers can send data directly, they need to agree on many things:

• What media types (audio, video, data) they support.
• Which codecs to use (e.g., Opus for audio, VP8 for video).
• Their network addresses and connection methods.

SDP is the standard way to exchange this crucial information.

WebRTC uses an "Offer/Answer" model for negotiation. One peer creates an SDP Offer, describing its capabilities. The other peer receives it and creates an SDP Answer, indicating its own capabilities and agreement.

This exchange happens via a signaling server, which we discussed in the previous lesson.

An SDP message is a collection of lines, each starting with a single lowercase letter followed by an equals sign (=). These lines are grouped into a session description and one or more media descriptions.

Common lines include: v=, o=, s=, t=, c=, m=, and a=.

Every SDP message begins with a version line and an origin line:

• v=0: Specifies the SDP protocol version. Always 0 for WebRTC.
• o= IN IP4
  : Describes the session's originator and identifier.

The <sess-id> and <sess-version> are unique numbers, and <address> is typically 127.0.0.1 for local generation.

Next, we find the session name and time description:

• s=-: The session name. Often just a hyphen for WebRTC.
• t=0 0: The session's active time. 0 0 means permanent/always active for WebRTC.

These lines provide basic context for the overall session.

The c= line provides connection data, specifically the IP address for the session. In WebRTC, this is often set to 0.0.0.0 or 127.0.0.1 as the actual peer-to-peer connection addresses are negotiated later via ICE candidates.

Example: c=IN IP4 0.0.0.0

The m= line is critical! It describes a specific media stream (audio, video, or data channel). An SDP can have multiple m= lines.

Format: m=

• <media>: audio, video, or application (for data channels).
• <port>: The port number (often 9 for WebRTC, actual port from ICE).
• <protocol>: E.g., UDP/TLS/RTP/SAVPF.
• <fmt list>: A list of supported payload types/codecs.

a= lines provide additional attributes for the session or a specific media stream. They are highly flexible and carry a lot of WebRTC-specific information.

• a=rtpmap: /: Maps payload types to codecs.
• a=fingerprint:...: TLS certificate fingerprint.
• a=sendrecv, a=sendonly, a=recvonly: Directionality of media.
• a=ice-ufrag, a=ice-pwd: Used for ICE connectivity checks.

Here's a simplified example of what an SDP offer might look like. Notice the different lines and their roles in describing the session and media capabilities.

Try running it to see the raw SDP text!

Which SDP line primarily describes the type of media (audio, video, or data), its associated port, transport protocol, and supported formats?

Great job! You've learned about SDP, the Session Description Protocol, which is fundamental to WebRTC.

• SDP is a text format for describing media and network capabilities.
• It's used in the Offer/Answer model to negotiate connections.
• Key lines include v=, o=, s=, t=, c=, m=, and a=.
• The m= line describes media, and a= lines provide rich attributes.

Next, we'll dive into ICE candidates, which help peers find the best network path!