RFC 8852: RTP Stream Identifier Source Description (SDES)

Status of This Memo

This is an Internet Standards Track document.¶

This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Further information on Internet Standards is available in Section 2 of RFC 7841.¶

Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at https://www.rfc-editor.org/info/rfc8852.¶

Copyright Notice

Copyright (c) 2021 IETF Trust and the persons identified as the document authors. All rights reserved.¶

This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.¶

1. Introduction

RTP sessions frequently consist of multiple streams, each of which is identified at any given time by its synchronization source (SSRC); however, the SSRC associated with a stream is not guaranteed to be stable over its lifetime. Within a session, these streams can be tagged with a number of identifiers, including CNAMEs and MediaStream Identification (MSID) [RFC8830]. Unfortunately, none of these have the proper ordinality to refer to an individual stream; all such identifiers can appear in more than one stream at a time. While approaches that use unique payload types (PTs) per stream have been used in some applications, this is a semantic overloading of that field, and one for which its size is inadequate: in moderately complex systems that use PT to uniquely identify every potential combination of codec configuration and unique stream, it is possible to simply run out of values.¶

To address this situation, we define a new RTCP Stream Identifier Source Description (SDES) identifier, RtpStreamId, that uniquely identifies a single RTP stream. A key motivator for defining this identifier is the ability to differentiate among different encodings of a single source stream that are sent simultaneously (i.e., simulcast). This need for unique identification extends to dependent streams (e.g., where layers used by a layered codec are transmitted on separate streams).¶

At the same time, when redundancy RTP streams are in use, we also need an identifier that connects such streams to the RTP stream for which they are providing redundancy. For this purpose, we define an additional SDES identifier, RepairedRtpStreamId. This identifier can appear only in packets associated with a redundancy RTP stream. They carry the same value as the RtpStreamId of the RTP stream that the redundant RTP stream is correcting.¶

2. Terminology

In this document, the terms "source stream", "RTP stream", "source RTP stream", "dependent stream", "received RTP stream", and "redundancy RTP stream" are used as defined in [RFC7656].¶

The following acronyms are also used:¶

3. Usage of RtpStreamId and RepairedRtpStreamId in RTP and RTCP

The RTP fixed header includes the payload type number and the SSRC values of the RTP stream. RTP defines how to demultiplex streams within an RTP session; however, in some use cases, applications need further identifiers in order to effectively map the individual RTP streams to their equivalent payload configurations in the SDP.¶

This specification defines two new RTCP SDES items [RFC3550]. The first item is "RtpStreamId", which is used to carry RTP stream identifiers within RTCP SDES packets. This makes it possible for a receiver to associate received RTP packets (identifying the RTP stream) with a media description having the format constraint specified. The second is "RepairedRtpStreamId", which can be used in redundancy RTP streams to indicate the RTP stream repaired by a redundancy RTP stream.¶

To be clear: the value carried in a RepairedRtpStreamId will always match the RtpStreamId value from another RTP stream in the same session. For example, if a source RTP stream is identified by RtpStreamId "A", then any redundancy RTP stream that repairs that source RTP stream will contain a RepairedRtpStreamId of "A" (if this mechanism is being used to perform such correlation). These redundant RTP streams may also contain their own unique RtpStreamId.¶

This specification also uses the RTP header extension for RTCP SDES items [RFC7941] to allow carrying RtpStreamId and RepairedRtpStreamId values in RTP packets. This allows correlation at stream startup, or after stream changes where the use of RTCP may not be sufficiently responsive. This speed of response is necessary since, in many cases, the stream cannot be properly processed until it can be identified.¶

RtpStreamId and RepairedRtpStreamId values are scoped by source identifier (e.g., CNAME) and by media session. When the media is multiplexed using the BUNDLE extension [RFC8843], these values are further scoped by their associated MID values. For example: an RtpStreamId of "1" may be present in the stream identified with a CNAME of "1234@example.com" and may also be present in a stream with a CNAME of "5678@example.org", and these would refer to different streams. Similarly, an RtpStreamId of "1" may be present with an MID of "A", and again with a MID of "B", and also refer to two different streams.¶

Note that the RepairedRtpStreamId mechanism is limited to indicating one repaired stream per redundancy stream. If systems require correlation for schemes in which a redundancy stream contains information used to repair more than one stream, they will have to use a more complex mechanism than the one defined in this specification.¶

As with all SDES items, RtpStreamId and RepairedRtpStreamId are limited to a total of 255 octets in length. RtpStreamId and RepairedRtpStreamId are constrained to contain only alphanumeric characters. For avoidance of doubt, the only allowed byte values for these IDs are decimal 48 through 57, 65 through 90, and 97 through 122.¶

     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |RtpStreamId=12 |     length    | RtpStreamId                 ...
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |Repaired...=13 |     length    | RepairRtpStreamId           ...
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

4. IANA Considerations

5. Security Considerations

Although the identifiers defined in this document are limited to be strictly alphanumeric, SDES items have the potential to carry any string. As a consequence, there exists a risk that they might carry privacy-sensitive information. Implementations need to take care when generating identifiers so that they do not contain information that can identify the user or allow for long-term tracking of the device. Following the generation recommendations in Section 3.3 will result in non-instance-specific labels, with only minor fingerprinting possibilities in the total number of used RtpStreamIds and RepairedRtpStreamIds.¶

Even if the SDES items are generated to convey as little information as possible, implementors are strongly encouraged to encrypt SDES items -- both in RTCP and RTP header extensions -- so as to preserve privacy against third parties.¶

As the SDES items are used for identification of the RTP streams for different application purposes, it is important that the intended values are received. An attacker, either a third party or malicious RTP middlebox, that removes or changes the values for these SDES items can severely impact the application. The impact can include failure to decode or display the media content of the RTP stream. It can also result in incorrectly attributing media content to identifiers of the media source, such as incorrectly identifying the speaker. To prevent this from occurring due to third-party attacks, integrity and source authentication is needed.¶

"Options for Securing RTP Sessions" [RFC7201] discusses options for how encryption, integrity, and source authentication can be accomplished.¶

Acknowledgements

Many thanks to Cullen Jennings, Magnus Westerlund, Colin Perkins, Jonathan Lennox, and Paul Kyzivat for review and input. Magnus Westerlund provided nearly all of the Security Considerations section.¶

Authors' Addresses