Duplication Grouping Semantics in the Session Description Protocol

The Real-time Transport Protocol (RTP) is widely used today for delivering IPTV traffic and other real-time multimedia sessions. Many of these applications support very large numbers of receivers and rely on intra-domain UDP/IP multicast for efficient distribution of traffic within the network. While this combination has proved successful, there does exist a weakness. As noted, packet loss is not avoidable, even in a carefully managed network. This loss might be due to congestion; it might also be a result of an unplanned outage caused by a flapping link, a link or interface failure, a software bug, or a maintenance person accidentally cutting the wrong fiber. Since UDP/IP flows do not provide any means for detecting loss and retransmitting packets, it is left up to the RTP layer and the applications to detect, and recover from, packet loss. One technique to recover from packet loss without incurring unbounded delay for all the receivers is to duplicate the packets and send them in separate redundant streams. Variations on this idea have been implemented and deployed today . explains how duplication can be achieved for RTP streams without breaking the RTP and RTP Control Protocol (RTCP) functionality. In this document, we describe the semantics needed in the Session Description Protocol (SDP) to support this technique.

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in .

Each "a=group" line is used to indicate an association relationship between the redundant streams. The streams included in one "a=group" line are called a "Duplication Group". Using the SDP Grouping Framework in , this document defines "DUP" as the grouping semantics for redundant streams. The "a=group:DUP" semantics MUST be used to group the redundant streams, except when the streams are specified in the same media description, i.e., in the same "m" line (see ). In an "a=group:DUP" line, the order of the listed redundant streams does not strictly indicate the order of transmission, although it is RECOMMENDED that the stream listed first be sent first, with the other stream(s) being the (time-delayed) duplicate(s).

defines an SDP media-level attribute, called "ssrc-group", for grouping the RTP streams that are SSRC multiplexed and carried in the same RTP session. The grouping is based on the SSRC identifiers. Since SSRC-multiplexed RTP streams are defined in the same "m" line, the "group" attribute cannot be used. This section explains how duplication is used with SSRC-multiplexed streams using the "ssrc-group" attribute . The semantics of "DUP" for the "ssrc-group" attribute are the same as the one defined for the "group" attribute, except that the SSRC identifiers are used to designate the duplication grouping associations: a=ssrc-group:DUP *(SP ssrc-id) . As above, while in an "a=ssrc-group:DUP" line, the order of the listed redundant streams does not necessarily indicate the order of transmission, but it is RECOMMENDED that the stream listed first be sent first, with the other stream(s) being the (time-delayed) duplicate(s).

When offering duplication grouping using SDP in an offer/answer model , the following considerations apply. A node that is receiving an offer from a sender may or may not understand line grouping. It is also possible that the node understands line grouping but does not understand the "DUP" semantics. From the viewpoint of the sender of the offer, these cases are indistinguishable. When a node is offered a session with the "DUP" grouping semantics but it does not support line grouping or the duplication grouping semantics, as per , the node responds to the offer either (1) with an answer that omits the grouping attribute or (2) with a refusal to the request (e.g., "488 Not Acceptable Here" or "606 Not Acceptable in SIP"). In the first case, the original sender of the offer must send a new offer without any duplication grouping. In the second case, if the sender of the offer still wishes to establish the session, it should retry the request with an offer without the duplication grouping. This behavior is specified in .

In this example, the redundant streams use the same IP destination address (232.252.0.1), but they are sourced from different addresses (198.51.100.1 and 198.51.100.2). Thus, the receiving host needs to join both source-specific multicast (SSM) sessions separately.

Note that in actual use, SSRC values, which are random 32-bit numbers, can be much larger than the ones shown in this example. Also, note that this SDP description does not use the "duplication-delay" attribute (defined in ) since the sender does not apply any delay between the redundant streams upon transmission. Alternatively, one MAY explicitly insert an "a=duplication-delay:0" line before the "a=mid:Ch1" line for informational purposes.

In this example, the redundant streams have different IP destination addresses. The example shows the same UDP port number and IP source address for each stream, but either or both could have been different for the two streams.

Optionally, one could be more explicit and insert an "a=duplication&nbhy;delay:0" line before the first "m" line.

In this example, the redundant streams have the same IP source and destination addresses (i.e., they are transmitted in the same SSM session). Due to the same source and destination addresses, the packets in both streams will be routed over the same path. To provide resiliency against packet loss, the duplicate of an original packet is transmitted 50 milliseconds (ms) later as indicated by the "duplication-delay" attribute (defined in ).

In general, the security considerations of apply to this document as well. There is a weak threat for the receiver that the duplication grouping can be modified to indicate relationships that do not exist. Such attacks might result in failure of the duplication mechanisms and/or mishandling of the media streams by the receivers. In order to avoid attacks of this sort, the SDP description needs to be integrity protected and provided with source authentication. This can, for example, be achieved on an end-to-end basis using S/MIME when the SDP is used in a signaling packet using MIME types (application/sdp). Alternatively, HTTPS or the authentication method in the Session Announcement Protocol (SAP) could be used as well. As for the confidentiality, if it is desired, it can be useful to use a secure, encrypted transport method to carry the SDP description.

This document registers the following semantics with IANA in the "Semantics for the "group" SDP Attribute" subregistry (under the "Session Description Protocol (SDP) Parameters" registry:

This document also registers the following semantics with IANA in the "Semantics for the "ssrc-group" SDP Attribute" subregistry under the "Session Description Protocol (SDP) Parameters" registry:

The authors would like to thank Colin Perkins, Bill Ver Steeg, Dave Oran, and Toerless Eckert for their input and suggestions.