Audio/Video Transport WG S. Wenger Internet Draft Y.-K. Wang Intended status: Standards track Nokia Expires: December 2008 T. Schierl Fraunhofer HHI A. Eleftheriadis Vidyo June 30, 2008 RTP Payload Format for SVC Video draft-ietf-avt-rtp-svc-12.txt Status of this Memo By submitting this Internet-Draft, each author represents that any applicable patent or other IPR claims of which he or she is aware have been or will be disclosed, and any of which he or she becomes aware will be disclosed, in accordance with Section 6 of BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet- Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html This Internet-Draft will expire on December 30, 2008. Copyright Notice Copyright (C) The IETF Trust (2008). Wenger, et al Expires December 30, 2008 [Page 1] Internet-Draft RTP Payload Format for SVC Video June 2008 Abstract This memo describes an RTP payload format for Scalable Video Coding (SVC) as defined in_Annex G of ITU-T Recommendation H.264, which is technically identical to Amendment 3 of ISO/IEC International Standard 14496-10. The RTP payload format allows for packetization of one or more H.264 Network Abstraction Layer (NAL) units in each RTP packet payload, supporting both single-session as well as multi- session streams. For single-session streams the packetization modes of RFC 3984 are used, whereas for multi-session streams four different packetization modes are defined in this memo. The payload format is backwards compatible to RFC 3984, and has wide applicability in conversational applications such as videoconferencing, Internet video streaming, and high bit-rate entertainment-quality video, among others. Table of Contents Status of this Memo ............................................ 1 Copyright Notice ............................................... 1 Abstract ....................................................... 2 Table of Contents .............................................. 2 1. Introduction ................................................ 4 2. Conventions ................................................. 6 3. Scope ....................................................... 7 4. Definitions and Abbreviations ............................... 7 4.1 Definitions ............................................. 7 4.1.1 Definitions from the SVC Specification ............. 7 4.1.2 Definitions Specific to This Memo .................. 9 4.2 Abbreviations .......................................... 12 5. The SVC Codec .............................................. 12 5.1 Overview ............................................... 12 5.2 Parameter Sets ......................................... 15 5.3 Network Abstraction Layer Units ........................ 16 6. RTP Payload Format ......................................... 20 6.1 Design Principles ...................................... 20 6.2 RTP Header Usage ....................................... 20 6.3 Common Structure of the RTP Payload Format ............. 20 6.4 NAL Unit Header Usage .................................. 20 6.5 Packetization Modes .................................... 22 6.5.1 Packetization Modes for Single-Source Transmission ...................................... 22 6.5.2 Packetization Modes for Multi-Source Transmission ...................................... 22 Wenger, et al Expires December 30, 2008 [Page 2] Internet-Draft RTP Payload Format for SVC Video June 2008 6.6 Aggregation Packets .................................... 25 6.7 Fragmentation Units (FUs) .............................. 25 6.8 Payload Content Scalability Information (PACSI) NAL Unit ............................................... 25 6.9 Non-Interleaved Multi-Time Aggregation Packets (NI-MTAPs) ............................................. 32 6.10 Decoding Order Number (DON) ........................... 34 6.10.1 Cross-Session DON (CS-DON) for Multi-Source Transmission ............................................ 35 7. Packetization Rules ........................................ 36 7.1 Packetization Rules for Multi-Source Transmission ...... 37 7.1.1 NI-T / NI-TC Packetization Rules .................. 38 7.1.2 NI-C / NI-TC Packetization Rules .................. 38 7.1.3 I-C Packetization Rules ........................... 40 7.1.4 Packetization Rules for Non-VCL NAL Units ......... 40 7.1.5 Packetization Rules for Prefix NAL Units .......... 40 8. De-Packetization Process ................................... 41 8.1 De-Packetization Process for Multi-Source Transmission . 41 8.1.1 Decoding Order Recovery for the NI-T and NI-TC Modes ....................................... 42 8.1.1.1 Informative Algorithm for NI-T Decoding Order Recovery within an Access Unit ............... 45 8.1.2 Decoding Order Recovery for the NI-C, NI-TC and I-C Modes ............................................. 48 9. Payload Format Parameters .................................. 50 9.1 Media Type Registration ................................ 50 9.2 SDP Parameters ......................................... 60 9.2.1 Mapping of Payload Type Parameters to SDP ......... 61 9.2.2 Usage with the SDP Offer/Answer Model.............. 61 9.2.3 Usage with Multi-Source Transmission .............. 66 9.2.4 Usage in Declarative Session Descriptions ......... 66 9.3 Examples ............................................... 67 9.3.1 Example for Offering A Single SVC Session ......... 67 9.3.2 Example for Offering Session Multiplexing ......... 68 9.4 Parameter Set Considerations ........................... 69 10. Security Considerations ................................... 69 11. Congestion Control ........................................ 69 12. IANA Consideration ........................................ 70 13. Informative Appendix: Application Examples................. 71 13.1 Introduction .......................................... 71 13.2 Layered Multicast ..................................... 71 13.3 Streaming of an SVC Scalable Stream ................... 72 13.4 Multicast to MANE, SVC Scalable Stream to Endpoint .... 73 13.5 Scenarios Currently Not Considered .................... 74 13.6 SSRC Multiplexing ..................................... 76 14. References ................................................ 76 14.1 Normative References................................... 76 Wenger, et al Expires December 30, 2008 [Page 3] Internet-Draft RTP Payload Format for SVC Video June 2008 14.2 Informative References................................. 77 15. Authors' Addresses......................................... 78 Intellectual Property Statement ............................... 79 Disclaimer of Validity......................................... 79 Copyright Statement............................................ 80 Acknowledgement................................................ 80 16. Open Issues................................................ 80 17. Changes Log................................................ 81 From draft-ietf-avt-rtp-svc-08 to draft-ietf-avt-rtp-svc-09 ... 81 From draft-ietf-avt-rtp-svc-09 to draft-ietf-avt-rtp-svc-10 ... 82 From draft-ietf-avt-rtp-svc-10 to draft-ietf-avt-rtp-svc-11 ... 83 From draft-ietf-avt-rtp-svc-11 to draft-ietf-avt-rtp-svc-12 ... 83 1. Introduction This memo specifies an RTP [RFC3550] payload format for the Scalable Video Coding (SVC) extension of the H.264/AVC video coding standard. SVC is specified in Amendment 3 to ISO/IEC 14496 Part 10 [MPEG4-10], and Annex G of ITU-T Rec. H.264/AVC [H.264]. SVC covers the entire application range of H.264/AVC, from low bitrate Internet streaming applications, to HDTV broadcasting, and even Digital Cinema that requires nearly lossless coding and hundreds of Mbps. The payload format specified in this memo is a backwards compatible enhancement to the H264/AVC payload format (H264, [RFC3984]), in which the specific features introduced by SVC are taken into account. It is assumed that the reader is familiar with the terminology and concepts defined in RFC 3984. SVC provides a coded representation of a video signal as a set of hierarchical components, composed of a base layer and one or more enhancement layers, as explained in Section 5 All data produced by an SVC encoder are structured in H.264 Network Abstraction Layer (NAL) units. This payload specification can only be used to carry the raw H.264 NAL unit stream over RTP, and not the byte stream format specified in Annex B of [H.264]. Depending on the packetization mode used, one or more than one NAL unit may be present in a single RTP packet. The base layer is, by design, compatible to H264, but may be formatted either according to RFC 3984 ("AVC base layer") or according to this memo ("SVC base layer"). Furthermore, the base layer may have multiple temporal components (i.e., supporting different frame rates). As a result, we distinguish the lowest temporal component ("T0") of the base layer (either AVC or SVC) as the starting point of the SVC bitstream Wenger, et al Expires December 30, 2008 [Page 4] Internet-Draft RTP Payload Format for SVC Video June 2008 hierarchy. The difference of an SVC base layer as compared to an AVC base layer is that additional NAL unit types may be present in the RTP stream in the SVC base layer case, which, however, are ignored by a receiver conforming to RFC 3984. This specification allows to encapsulate in a given RTP stream NAL units belonging to either: o the T0 AVC base layer or the T0 SVC base layer only; o one or more enhancement layers; or o the T0 SVC base layer, and one or more enhancement layers. Furthermore, this specification allows the packetization of SVC data for either single-source or multi-source transmission. In the case of single-source transmission (SST) all SVC data are carried in a single RTP session with the same SSRC. In the case of Multi-Source Transmission (MST), two or more RTP sessions are used to carry the SVC data, using distinct SSRC's, in accordance with the packetization modes defined in this memo and in RFC 3984. Each RTP session is associated with one RTP stream, which MAY carry one or more layers, structured according to one of the three cases indicated above. When MST is not used, the following applies. o When an H.264/AVC compatible subset of the SVC base layer is transmitted, the subset SHOULD be carried in one RTP stream that MUST be encapsulated according to RFC 3984. This way, an RFC 3984 receiver will be able to receive the H.264/AVC compatible bitstream subset. o When a set of layers including one or more SVC enhancement layers is transmitted, the set SHOULD be carried in one RTP stream that SHOULD be encapsulated according to this memo. When MST is used, this memo defines four different packetization modes. The modes differ depending on if the SVC data are allowed to be interleaved, i.e., to be transmitted in an order different than the intended decoding order, and they also differ in the mechanisms provided in order to recover the correct decoding order of the NAL units across the multiple RTP sessions. These four MST modes re-use the packetization modes introduced in RFC 3984 for the packetization of NAL units in each of their individual RTP sessions. Wenger, et al Expires December 30, 2008 [Page 5] Internet-Draft RTP Payload Format for SVC Video June 2008 MST SHOULD be used in a multicast session when different receivers may request different layers of the scalable bitstream. An operation point for an SVC bit stream, as defined in this memo, corresponds to a set of layers that together conform to one of the profiles defined in Annex A or G of [H.264] and, when decoded, offer a representation of the original video at a certain fidelity. The number of streams used in MST SHOULD be at least equal to the number of operation points that may be requested by the receivers. Depending on the application, this may result in each layer being carried in its own RTP session, or in having multipe layers encapsulated within one RTP session. Informative note: Layered multicast is a term commonly used to describe the application where multicast is used to transmit layered or scalable data that has been encapsulated into more than one RTP session. This application allows different receivers in the multicast session to receive different operation points of the scalable bitstream. Layered multicast, among other application examples, is discussed in more detail in Section 13.2 This RTP payload specification is designed to be unaware of the NAL unit payload defined in [H.264]. Similar to RFC 3984, this memo introduces two new NAL unit types, using unit type numbers from the space explicitly left unspecified in [H.264] and not used in RFC 3984. When the single NAL unit packetization mode is used, where one NAL unit always corresponds to one RTP packet, the NAL unit header defined in [H.264] co-serves as the payload header of this RTP payload format. In this case, the payload of the NAL unit follows immediately. In all other modes data from multiple NAL units may be present in an RTP packet, either through nesting (a NAL unit is contained in another one) or serialization (NAL units appear in sequence in an RTP packet). This memo also also defines signaling support for SVC, including a new media subtype name (H264-SVC). 2. Conventions The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14, RFC 2119 [RFC2119]. This specification uses the notion of setting and clearing a bit when bit fields are handled. Setting a bit is the same as assigning Wenger, et al Expires December 30, 2008 [Page 6] Internet-Draft RTP Payload Format for SVC Video June 2008 that bit the value of 1 (On). Clearing a bit is the same as assigning that bit the value of 0 (Off). 3. Scope o The scalability features that SVC adds to the H.264 specification enable several system-level functionalities related to the ability of a system to adapt the signal to different system conditions with no or minimal processing. The adaptation relates both to the capabilities of potentially heterogenous receivers (screen resolution, processing speed, etc.), as well as differing or time-varying network conditions. The adaptation can be performed at the source, the destination, or in intermediate media-aware network elements. This payload specification exposes these system-level functionalities so that system designers can take direct advantage of these features. The likely applications of this specification are in the IP-based multimedia communication fields, including conversational multimedia, video telephony or video conferencing, Internet streaming, and IPTV. 4. Definitions and Abbreviations 4.1 Definitions 4.1.1 Definitions from the SVC Specification This document uses the terms and definitions of [H.264]. The following terms are relevant to this memo, and their definitions are copied here from [H.264] for convenience. access unit: A set of NAL units always containing exactly one primary coded picture. In addition to the primary coded picture, an access unit may also contain one or more redundant coded pictures, one auxiliary coded picture, or other NAL units not containing slices or slice data partitions of a coded picture. The decoding of an access unit always results in a decoded picture. base layer: A bitstream subset that contains all the NAL units with the nal_unit_type syntax element equal to 1 or 5 of the bitstream and does not contain any NAL unit with the nal_unit_type syntax element equal to 14, 15, or 20 and conforms to one or more of the profiles specified in Annex A of [H.264]. base quality layer representation: The layer representation of the target dependency representation of an access unit that is associated with the quality_id syntax element equal to 0. Wenger, et al Expires December 30, 2008 [Page 7] Internet-Draft RTP Payload Format for SVC Video June 2008 coded video sequence: A sequence of access units that consists, in decoding order, of an IDR access unit followed by zero or more non-IDR access units including all subsequent access units up to but not including any subsequent IDR access unit. dependency representation: A subset of VCL NAL units within an access unit that are associated with the same value of the dependency_id syntax element, which is provided as part of the NAL unit header or by an associated prefix NAL unit. A dependency representation consist of one or more layer representations. IDR access unit: An access unit in which the primary coded picture is an IDR picture. IDR picture: A coded picture in which all slices of the target dependency representation within the access unit are I or EI slices that causes the decoding process to mark all reference pictures as "unused for reference" immediately after decoding the IDR picture. After the decoding of an IDR picture all following coded pictures in decoding order can be decoded without inter prediction from any picture decoded prior to the IDR picture. The first picture of each coded video sequence is an IDR picture. layer representation: A subset of VCL NAL units within an access unit that are associated with the same values of the dependency_id and quality_id syntax elements, which are provided as part of the VCL NAL unit header or by an associated prefix NAL unit. One or more layer representations represent a dependency representation. prefix NAL unit: A NAL unit with nal_unit_type equal to 14 that immediately precedes in decoding order a NAL unit with nal_unit_type equal to 1, 5, or 12. The NAL unit that immediately succeeds in decoding order the prefix NAL unit is referred to as the associated NAL unit. The prefix NAL unit contains data associated with the associated NAL unit, which are considered to be part of the associated NAL unit. reference base picture: A reference picture that is obtained by decoding a base quality layer representation with the nal_ref_idc syntax element not equal to 0 and the store_ref_base_pic_flag syntax element equal to 1 of an access unit and all layer representations of the access unit that are referred to by inter- layer prediction of the base quality layer representation. A reference base picture is not an output of the decoding process, but the samples of a reference base picture may be used for inter Wenger, et al Expires December 30, 2008 [Page 8] Internet-Draft RTP Payload Format for SVC Video June 2008 prediction in the decoding process of subsequent pictures in decoding order. Reference base picture is a collective term for a reference base field or a reference base frame. scalable bitstream: A bitstream with the property that one or more bitstream subsets that are not identical to the scalable bitstream form another bitstream that conforms to the SVC specification[SVC]. target dependency representation: The dependency representation of an access unit that is associated with the largest value of the dependency_id syntax element for all dependency representations of the access unit. target layer representation: The layer representation of the target dependency representation of an access unit that is associated with the largest value of the quality_id syntax element for all layer representations of the target dependency representation of the access unit. 4.1.2 Definitions Specific to This Memo anchor layer representation: An anchor layer representation is such a layer representation that, if decoding of the operation point corresponding to the layer starts from the access unit containing this layer representation, all the following layer representations of the layer, in output order, can be correctly decoded. An anchor layer representation is a random access point to the layer the anchor layer representation belongs to. However, some layer representations, succeeding an anchor layer representation in decoding order but preceding the anchor layer representation in output order, may refer to earlier layer representations for inter prediction, and hence may not be correctly decoded if random access is performed at the anchor layer representation. AVC base layer: The subset of the SVC base layer in which all prefix NAL units (type 14) are removed. Note that this is equivalent to the term "base layer" as defined in Annex G of [H.264]. base RTP session: When multi-source transmission is used, the RTP session that carries the RTP stream containing the T0 AVC base layer or the T0 SVC base layer, and zero or more enhancement layers. This RTP session does not depend on any other RTP session as indicated by mechanisms defined in [I-D.ietf-mmusic- Wenger, et al Expires December 30, 2008 [Page 9] Internet-Draft RTP Payload Format for SVC Video June 2008 decoding-dependency]. The base RTP session may carry NAL units of NAL unit type equal to 14 and 15. effective NAL unit timestamp: The value that the RTP timestamp would have if the particular NAL unit was transported in its own RTP packet. (The NAL unit time is different than that actual RTP timestamp of the packet containing the particular NAL unit in the case of MTAPs.) enhancement RTP session: When multi-source transmission is used, an RTP session that is not the base RTP session. An enhancement RTP session typically contains an RTP stream that depends on at least one other RTP session as indicated by mechanisms defined in [I-D.ietf-mmusic-decoding-dependency]. A lower RTP session to an enhancement RTP session is an RTP session which the enhancement RTP session depends on. The lowest RTP session for a receiver is the RTP session that does not depend on any other RTP session received by the receiver. The highest RTP session for a receiver is the RTP session which no other RTP session received by the receiver depends on. cross-session decoding order number (CS-DON): A derived variable indicating NAL unit decoding order number over all NAL units within all the session-multiplexed RTP sessions that carry the same SVC bitstream. enhancement layer: A layer in which at least one of the values of dependency_id or quality_id is higher than 0, or a layer in which none of the NAL units is associated with the value of temporal_id equal to 0. An operation point constructed using the maximum temporal_id, dependency_id, and quality_id values associated with an enhancement layer may or may not conform to one or more of the profiles specified in Annex A of [H.264]. H.264/AVC compatible: A biststream subset that conforms to one or more of the profiles specified in Annex A of [H.264]. intra layer representation: A layer representation that contains only slices that use intra prediction, and hence do not refer to any earlier layer representation in decoding order in the same layer. Note that in [SVC] intra prediction includes intra-layer intra prediction as well as inter-layer intra prediction. layer: A bistream subset in which all NAL units of type 1, 5, 12, 14, or 20 have the same values of dependency_id and quality_id, either directly through their NAL unit header (for NAL units of type 14 or 20) or through association to a prefix (type 14) NAL Wenger, et al Expires December 30, 2008 [Page 10] Internet-Draft RTP Payload Format for SVC Video June 2008 unit (for NAL unit types 1, 5, or 12). A layer may contain NAL units associated with more than one values of temporal_id. multi-source transmission: This specifies that the SVC bitstream is distributed across multiple RTP sessions, with each stream having a distinct SSRC, and consequently its own timestamp and sequence number spaces. Those multiple streams can be associated using the RTCP CNAME, or explicit signalling of the SSRC used. [Ed. (AE): Is the single transport connection mode supported? It does not appear to, as seen by the definitions of base and enhancement RTP sessions, and the rest of the text. I modified the definition so that it is not allowed.] Dependency between RTP sessions MUST be signaled according to [I-D.ietf-mmusic-decoding- dependency] and this memo. operation point: An operation point is identified by a set of values of temporal_id, dependency_id, and quality_id. A bistream corresponding to an operation point can be constructed by removing all NAL units associated with a higher value of dependency_id, and all NAL units associated with the same value of dependency_id but higher values of quality_id or temporal_id. Additional NAL units may be removed (with lower dependency_id or same dependency_id but lower quality_id) if they are not required for decoding the bitstream at the particular operation point. An operation point bitstream conforms to at least one of the profiles defined in Annex A or Annex G of [H.264], and offers a representation of the original video signal at a certain fidelity. [Ed.Note(YkW): Need to check whether a bitstream subset with those additional NAL units removed is a conforming bitstream.] operation point representation: The set of all NAL units of an operation point within the same access unit. RTP packet stream: A sequence of RTP packets with increasing sequence numbers (except for wrap-around), identical PT and identical SSRC (Synchronization Source), carried in one RTP session. Within the scope of this memo, one RTP packet stream is utilized to transport one or more layers. SVC base layer: The layer that includes all NAL units associated with dependency_id and quality_id values both equal to 0, including prefix NAL units (NAL unit type 14). SVC enhancement layer: A layer in which at least one of the values of dependency_id or quality_id is higher than 0. An operation point constructed using the maximum dependency_id and Wenger, et al Expires December 30, 2008 [Page 11] Internet-Draft RTP Payload Format for SVC Video June 2008 quality_id values and any temporal_id value associated with an SVC enhancement layer does not conform to any of the profiles specified in Annex A of [H.264]. SVC NAL unit: A NAL unit of NAL unit type 14, 15, or 20 as specified in Annex G of [H.264]. SVC NAL unit header: A four-byte header resulting from the addition of a three-byte SVC-specific header extension added in NAL unit types 14, 15 and 20. SVC RTP session: Either the base RTP session or an enhancement RTP session. T0 AVC base layer: A subset of the AVC base layer constructed by removing all VCL NAL units associated with temporal_id values higher than 0. T0 SVC base layer: A subset of the SVC base layer constructed by removing all VCL NAL units associated with temporal_id values higher than 0 as well as their associated prefix NAL units. 4.2 Abbreviations In addition to the abbreviations defined in [RFC3984], the following abbrevations are used in this memo. CGS: Coarse-Grain Scalability CS-DON: Cross-Session Decoding Order Number ETS: Effective Timestamp (of a NAL unit) MGS: Medium-Grain Scalability MST: Multi-Source Transmission PACSI: Payload Content Scalability Information SST: Single-Source Transmission SNR: Signal-to-Noise Ratio SVC: Scalable Video Coding 5. The SVC Codec 5.1 Overview SVC [H.264]defines a coded video representation in which a given bitstream offers representations of the source material at different levels of fidelity (hence the term "scalable"). Scalable video coding bitstreams, or scalable bitstreams, are constructed in a Wenger, et al Expires December 30, 2008 [Page 12] Internet-Draft RTP Payload Format for SVC Video June 2008 pyramidal fashion: the coding process creates bitstream components that improve the fidelity of hierarchically lower components. The fidelity dimensions offered by SVC are spatial (picture size), quality (or Signal-to-Noise Ratio - SNR), as well as temporal (pictures per second). Bitstream components associated with a given level of spatial, quality, and temporal fidelity are identified using corresponding parameters in the bitstream: dependency_id, quality_id, and temporal_id (see also Section 5.3). The fidelity identifiers have integer values, where higher values designate components that are higher in the hierarchy. It is noted that SVC offers significant flexibility in terms of how an encoder may choose to structure the dependencies between the various components. Decoding of a particular component requires the availability of all the components it depends upon, either directly, or indirectly. An operation point of an SVC bitstream consists of the bistream components required to be able to decode a particular dependency_id, quality_id, and temporal_id combination. SVC maintains the bitstream organization introduced in H.264/AVC. Specifically, all bitstream components are encapsulated in Network Abstraction Layer (NAL) units which are organized as Access Units (AU). An AU is associated with a single sampling instance in time. A subset of the NAL unit types correspond to the Video Coding Layer (VCL), and contain the coded picture data associated with the source content. Non-VLC NAL units carry ancillary data that may be necessary for decoding (e.g., parameter sets as explained below), or that facilitate certain system operations but are not needed by the decoding process itself. Coded picture data at the various fidelity dimensions are organized in slices. Within one AU, a coded picture of an operation point consists of all the coded slices required for decoding up to the particular combination of dependency_id and quality_id values at the time instance corresponding to the AU. The NAL encapsulates each slice generated by the VCL into one or more NAL units. RFC 3984 provides a more in-depth discussion of the NAL unit concept. SVC specifies the decoding order of NAL units. It is noted that the concept of temporal scalability is already present in H.264/AVC, as profiles defined in Annex A of [H.264] already support it. Specifically, in [H.264] sub-sequences have been introduced in order to allow optional use of temporal layers. SVC extends this approach by exposing the temporal scalability information using the temporal_id parameter, alongside (and unified with) the dependency_id and quality_id values that are used for spatial and quality scalability. For coded picture data defined in Annex G of [H.264] this is accomplished by using a new type of NAL unit where the fidelity parameters are part of its header. For Wenger, et al Expires December 30, 2008 [Page 13] Internet-Draft RTP Payload Format for SVC Video June 2008 coded picture data that follow H.264/AVC, and to ensure compatibility with existing H.264/AVC receivers, a new type of "prefix" NAL unit has been defined to carry this header information. This prefix NAL unit type is among those ignored by H.264/AVC receivers as explained in [RFC3984]. Within an AU, the VCL NAL units associated with a given dependency_id and quality_id are referred to as a "layer representation". The layer representation corresponding to the lowest values of dependency_id and quality_id (i.e., zero) is referred to as the base layer representation and is compliant by design to H.264/AVC. The set of VCL and associated non-VCL NAL units across all AUs in a bitstream associated with a particular combination of values of dependency_id and quality_id, and regardless of the value of temporal_id, is conceptually a scalable layer. Due to the backwards compatibility with H.264/AVC, it is important to differentiate, however, whether or not SVC-specific NAL units are present in a given bitstream or not. This is particularly important for the lowest fidelity values in terms of dependency_id and quality_id (zero for both), as the corresponding VCL data are compliant to H.264/AVC, and may or may not be accompanied by associated prefix NAL units. This memo therefore uses the term "AVC base layer" to designate the layer that contains only H.264/AVC VCL NAL units, and "SVC base layer" to designate the same layer but with the addition of the associated SVC prefix NAL units. Note that the SVC specification uses the term "base layer" for what in this memo will be referred to as "AVC base layer". Similarly, it is also important to be able to differentiate, within a layer, the temporal fidelity components it contains. This memo uses the term "T0" to indicate, within a particular layer, the subset that contains the NAL units associated with temporal_id equal to 0. The term "layer" is used in various contexts in this memo. For example, in the terms "Video Coding Layer" and "Network Abstraction Layer" it refers to conceptual organization levels. When referring to bitstream syntax elements such as block layer or macroblock layer, it refers to hierarchical bitstream structure levels. When used in the context of bitstream scalability, e.g., "AVC base layer", it refers to a level of representation fidelity of the source signal with a specific set of NAL units included. The correct interpretation is supported by providing the appropriate context. SNR scalability in SVC is offered in two different ways. In what is called Coarse-Grained Scalability (CGS), scalability is provided by including or excluding a complete layer when decoding a particular bitstream. In contrast, in Medium-Grained Scalability (MGS), Wenger, et al Expires December 30, 2008 [Page 14] Internet-Draft RTP Payload Format for SVC Video June 2008 scalability is provided by selectively omitting the decoding of specific NAL units belonging to MGS layers. The selection of the NAL units to omit can be based on fixed length fields present in the NAL unit header. 5.2 Parameter Sets The parameter set concept is defined in [H.264]. Please refer to Section 1.2 of RFC 3984 for more details. SVC introduces a new type of sequence parameter set, referred to as a subset sequence parameter set [H.264]. Subset sequence parameter sets have NAL unit type equal to 15, which is different from the NAL unit type value (7) of sequence parameter sets. VCL NAL units of NAL unit type 1 to 5 must only (indirectly) refer to sequence parameter sets, while VCL NAL units of NAL unit type 20 must only (indirectly) refer to subset sequence parameter sets. The references are indirect because VCL NAL units refer to picture parameter sets (in their slice header), which in turn refer to sequence parameter sets. Subset sequence parameter sets use a separate identifier value space than sequence parameter sets. An overview of the NAL unit and packet types used in this memo can be found in Table 1 in Section 5.3. In SVC, coded picture data from different layers may use the same or different sequence and picture parameter sets. At any time instant during the decoding process there may be one active sequence parameter set (for the layer representation with the highest value of (dependency_id * 16 + quality_id)) and one or more active layer SVC sequence parameter set(s) (for layer representations with lower values of (dependency_id * 16 + quality_id)). The active sequence parameter set or an active layer SVC sequence parameter set remains unchanged throughout a coded video sequence in the scalable layer in which the active sequence parameter set or active layer SVC sequence parameter set is referred to. This means that the referred sequence parameter set or subset sequence parameter set can only change at IDR access units for any layer. At any time instant during the decoding process there may be one active picture parameter set (for the layer representation with the highest value of (dependency_id * 16 + quality_id)) and one or more active layer picture parameter set(s) (for layer representations with lower values of (dependency_id * 16 + quality_id)). The active picture parameter set or an active layer picture parameter set remains unchanged throughout a layer representation in which the active picture parameter set or active layer picture parameter set is referred to, but may change from one AU to the next. Wenger, et al Expires December 30, 2008 [Page 15] Internet-Draft RTP Payload Format for SVC Video June 2008 5.3 Network Abstraction Layer Units The NAL unit organization is central to [ H.264], RFC 3984, as well as this memo. In addition to the NAL unit types defined originally for H.264/AVC, [H.264]introduces two new NAL unit types for SVC (among others): SVC VCL NAL units ("slice in scalable extension", type 20), and prefix NAL units (type 14). SVC VCL NAL units encapsulate VCL data as defined in Annex G of [H.264]. The prefix NAL unithas no payload of its own, and instead includes descriptive information of the associated H.264/AVC VCL NAL unit (type 1 or 5) that immediately follows the prefix NAL unit. In addition to the NAL unit types introduced for packetization purposes in RFC 2984, this memo also introduces two new NAL unit types to facilitate packetization (PACSI and NI-MTAP, specified in detail later on). The following table gives an overview of NAL unit and packet types used in this memo and also provides references to the appropriate document and section where their use is defined. Table 1. Summary of NAL unit and packet types used in this memo Type Description Definition in: [RC3984] / this memo -------------------------------------------------------------------- 0 unspecified - / - 1-23 NAL unit per [H.264]/Single NAL unit packet 5.2 / 6.3 14 Prefix NAL unit per [SVC] - / 5.1 15 Subset sequence parameter set per [SVC] - / 5.2 20 Slice in scalable extensions per [SVC] - / 5.3 24 Single-time aggregation packet (STAP-A) 5.7.1 / 6.6 25 Single-time aggregation packet (STAP-B) 5.7.1 / 6.6 26 Multi-time aggregation packet (MTAP16) 5.7.2 / 6.6 27 Multi-time aggregation packet (MTAP24) 5.7.2 / 6.66.7 28 Fragmentation unit (FU-A) 5.8 / 6.7 Wenger, et al Expires December 30, 2008 [Page 16] Internet-Draft RTP Payload Format for SVC Video June 2008 29 Fragmentation unit with DON (FU-B) 5.8 / 6.7 30 Payload Content Scalability Info. (PACSI) - / 6.8 31 unspecified - / - SVC extends the one-byte H.264/AVC NAL unit header by three additional octets for NAL units of type 14 and 20. The header indicates the type of the NAL unit, the (potential) presence of bit errors or syntax violations in the NAL unit payload, information regarding the relative importance of the NAL unit for the decoding process, the layer identification information, and other fields as discussed below. The syntax and semantics of the NAL unit header are specified in [H.264], but the essential properties of the NAL unit header are summarized below for convenience. The first byte of the NAL unit header has the following format (the bit fields are the same as defined for the one-byte H.264/AVC NAL unit header, while the semantics of some fields have changed slightly, in a backward compatible way): +---------------+ |0|1|2|3|4|5|6|7| +-+-+-+-+-+-+-+-+ |F|NRI| Type | +---------------+ The semantics of the components of the NAL unit type octet, as specified in the H.264 specification, are described briefly below. F: 1 bit forbidden_zero_bit. H.264/AVC declares a value of 1 as a syntax violation. NRI: 2 bits nal_ref_idc. A value of '00' (in binary form) indicates that the content of the NAL unit is not used to reconstruct reference pictures for future prediction. Such NAL units can be discarded without risking the integrity of the reference pictures in the same layer. A value greater than '00' indicates that the decoding of the NAL unit is required to maintain the integrity of reference pictures in the same layer, or that the NAL unit contains parameter sets. Wenger, et al Expires December 30, 2008 [Page 17] Internet-Draft RTP Payload Format for SVC Video June 2008 Type: 5 bits nal_unit_type. This component specifies the NAL unit type as defined in Table 7-1 of [H.264], and later within this memo. For a reference of all currently defined NAL unit types and their semantics, please refer to Section 7.4.1 in [H.264]. In H.264/AVC, NAL unit types 14, 15 and 20 are reserved for future extensions. SVC uses these three NAL unit types as follows: NAL unit type 14 is used for prefix NAL unit, NAL unit type 15 is used for subset sequence parameter set, and NAL unit type 20 is used for coded slice data in scalable extension (see Section 7.4.1 in [H.264]). NAL unit types 14 and 20 indicate the presence of three additional octets in the NAL unit header, as shown below. +---------------+---------------+---------------+ |0|1|2|3|4|5|6|7|0|1|2|3|4|5|6|7|0|1|2|3|4|5|6|7| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |R|I| PRID |N| DID | QID | TID |U|D|O| RR| +---------------+---------------+---------------+ R: 1 bit reserved_one_bit. Reserved bit for future extension. R MUST be equal to 1. Receivers SHOULD ignore the value of R. I: 1 bit idr_flag. This component specifies whether the layer representation is an instantaneous decoding refresh (IDR) layer representation (when equal to 1) or not (when equal to 0). PRID: 6 bits priority_id. This flag specifies a priority identifier for the NAL unit. A lower value of PRID indicates a higher priority. N: 1 bit no_inter_layer_pred_flag. This flag specifies, when present in a coded slice NAL unit, whether inter-layer prediction may be used for decoding the coded slice (when equal to 1) or not (when equal to 0). DID: 3 bits dependency_id. This component indicates the inter-layer coding dependency level of a layer representation. At any access unit, a layer representation with a given dependency_id may be used for inter-layer prediction for coding of a layer representation with a higher dependency_id, while a layer representation with a given Wenger, et al Expires December 30, 2008 [Page 18] Internet-Draft RTP Payload Format for SVC Video June 2008 dependency_id shall not be used for inter-layer prediction for coding of a layer representation with a lower dependency_id. QID: 4 bits quality_id. This component indicates the quality level of an MGS layer representation. At any access unit and for identical dependency_id values, a layer representation with quality_id equal to ql uses a layer representation with quality_id equal to ql-1 for inter-layer prediction. TID: 3 bits temporal_id. This component indicates the temporal level of a layer representation. The temporal_id is associated with the frame rate, with lower values of _temporal_id corresponding to lower frame rates. A layer representation at a given temporal_id typically depends on layer representations with lower temporal_id values, but it never depends on layer representations with higher temporal_id values. U: 1 bit use_ref_base_pic_flag. A value of 1 indicates that only reference base pictures are used during the inter prediction process. A value of 0 indicates that the reference base pictures are not used during the inter prediction process. D: 1 bit discardable_flag. A value of 1 indicates that the current NAL unit is not used for decoding NAL units with values of dependency_id higher than the one of the current NAL unit, in the current and all subsequent access units. Such NAL units can be discarded without risking the integrity of layers with higher dependency_id values. discardable_flag equal to 0 indicates that the decoding of the NAL unit is required to maintain the integrity of layers with higher dependency_id. O: 1 bit output_flag: Affects the decoded picture output process as defined in Annex C of [H.264]. RR: 2 bits reserved_three_2bits. Reserved bits for future extension. RR MUST be equal to '11' (in binary form). Receivers SHOULD ignore the value of RR. This specification extends the semantics of F, NRI, I, PRID, DID, QID, TID, U, and D per [H.264] as described in Section 6.4. Wenger, et al Expires December 30, 2008 [Page 19] Internet-Draft RTP Payload Format for SVC Video June 2008 6. RTP Payload Format 6.1Design Principles The following design principles have been observed: o Backward compatibility with [RFC3984] wherever possible. o The SVC base layer or any H.264/AVC compatible subset of the SVC base layer, when transmitted in its own RTP stream, MUST be encapsulated using [RFC3984]. This ensures that such an RTP stream can be understood by RFC 3984 receivers. o Media-Aware Network Elements (MANEs) as defined in [RFC3984] are signaling-aware and rely on signaling information. MANEs have state. o MANEs can aggregate multiple RTP streams, possibly from multiple RTP sessions. o MANEs can perform media-aware stream thinning (selective elimination of packets or portions thereof). By using the payload header information identifying Layers within an RTP session, MANEs are able to remove packets from the incoming RTP packet stream. This implies rewriting the RTP headers of the outgoing packet stream and rewriting of RTCP Receiver Reports. 6.2 RTP Header Usage Please see Section 5.1 of [RFC3984]. 6.3 Common Structure of the RTP Payload Format Please see Section 5.2 of [RFC3984]. 6.4 NAL Unit Header Usage The structure and semantics of the NAL unit header were introduced in Section 5.3. This section specifies the semantics of F, NRI, I, PRID, DID, QID, TID, U, and D according to this specification. The semantics of F specified in Section 5.3 of [RFC3984] also apply in this memo. For NRI, for a bitstream conforming to one of the profiles defined in Annex A of [H.264] and transported using [RFC3984], the semantics specified in Section 5.3 of [RFC3984] apply, i.e., NRI also Wenger, et al Expires December 30, 2008 [Page 20] Internet-Draft RTP Payload Format for SVC Video June 2008 indicates the relative importance of NAL units. For a bitstream conforming to one of the profiles defined in Annex G of [H.264] and transported using this memo, in addition to the semantics specified in Annex G of [H.264], NRI also indicates the relative importance of NAL units within a layer. For I, in addition to the semantics specified in Annex G of [H.264], according to this memo, MANEs MAY use this information to protect NAL units with I equal to 1 better than NAL units with I equal to 0. MANEs MAY also utilize information of NAL units with I equal to 1 to decide when to forward more packets for an RTP packet stream. For example, when it is sensed that spatial layer switching has happened such that the operation point has changed to a higher value of DID, MANEs MAY start to forward NAL units with the higher value of DID only after forwarding a NAL unit with I equal to 1 with the higher value of DID. Note that, in the context of this section, "protecting a NAL unit" means any RTP or network transport mechanism that could improve the probability of success delivery of the packet conveying the NAL unit, including applying a QoS-enabled network, Forward Error Correction (FEC), retransmissions, and advanced scheduling behavior, whenever possible. For PRID, the semantics specified in Annex G of [H.264] applies. Note, that MANEs implementing unequal error protection MAY use this information to protect NAL units with smaller PRID values better than those with larger PRID values, for example by including only the more important NAL units in an FEC protection mechanism. The importance for the decoding process decreases as the PRID value increases. For DID, QID, TID, in addition to the semantics specified in Annex G of [H.264], according to this memo, values of DID, QID, or TID indicate the relative importance in their respective dimension. A lower value of DID, QID, or TID indicates a higher importance if the other two components are identical. MANEs MAY use this information to protect more important NAL units better than less important NAL units. For U, in addition to the semantics specified in Annex G of [H.264], according to this memo, MANEs MAY use this information to protect NAL units with U equal to 1 better than NAL units with U equal to 0. For D, in addition to the semantics specified in Annex G of [H.264], according to this memo, MANEs MAY use this information to determine whether a given NAL unit is required for successfully decoding a Wenger, et al Expires December 30, 2008 [Page 21] Internet-Draft RTP Payload Format for SVC Video June 2008 certain Operation Point of the SVC bitstream, hence to decide whether to forward the NAL unit. 6.5 Packetization Modes 6.5.1Packetization Modes for Single-Source Transmission Section 5.4 of RFC 3984 applies when using single-source transmission. The packetization modes specified in Section 5.4 of RFC 3984 are also referred to as session-specific packetization modes. 6.5.2 Packetization Modes for Multi-Source Transmission When multi-source transmission (MST) is used this memo specifies four cases of MST packetization modes: o Non-interleaved timestamp based mode (NI-T); o Non-interleaved cross-session decoding order number (CS-DON) based mode (NI-C); o Non-interleaved combined timestamp and CS-DON mode (NI-TC); and o Interleaved CS-DON (I-C) mode. These four modes differ in two ways. First, they differ in terms of if they require that the NAL units are transmited in NAL unit decoding order (non-interleaved) or if they allow them to be transmitted in an arbitrary order (interleaved). Second, they differ in the mechanisms they provide in order to recover the correct decoding order of the NAL units across all RTP sessions involved. The NI-T, NI-C, and NI-TC modes do not allow interleaving, and are thus targeted for systems that require relatively low end-to-end latency, e.g. conversational systems. The I-C mode allows interleaving and is thus targeted for systems that do not require very low end-to-end latency. The NI-T and NI-TC modes use timestamps to recover the decoding order of NAL units, whereas NI-TC, NI-C, and I-C all use the CS-DON mechanism (explained later on) to do so. Note that the NI-TC mode uses both timestamps and the CS-DON method; receivers in this case may use either method for performing decoding order recovery. Wenger, et al Expires December 30, 2008 [Page 22] Internet-Draft RTP Payload Format for SVC Video June 2008 The MST packetization mode in use MAY be signaled by the value of the OPTIONAL pmode media type parameter or by external means. When the value of pmode is equal to "NI-T", the NI-T mode MUST be used. When the value of pmode is equal to "NI-C", the NI-C mode MUST be used. When the value of pmode is equal to "NI-TC" or pmode is not present, the NI-TC mode MUST be used. When the value of pmode is equal to "I-C", the I-C mode MUST be used. [Ed.Note(YkW): There MAY be at most one global pmode present in the SDP common for all the multiplexed RTP sessions. It is also possible to have pmode session-specific in the SDP, but then all the multiplexed sessions MUST have the same value of this parameter. When pmode is not present, the NI-TC mode is implied.] The used MST packetization mode governs which session-specific packetization modes are allowed in the associated RTP sessions, which in turn govern which NAL unit types are allowed as RTP payloads. Table 2 summarizes the allowed session-specific packetization modes for the NI-T, NI-C and NI-TC packetization modes. Table 3 summarizes the allowed session-specific packetization modes for the I-C packetization mode. Table 2 Summary of allowed session-specific packetization modes for the NI-T, NI-C and NI-TC packetization modes (yes = allowed, no = disallowed) Session-Specific Mode Base Session Enhancement Session ---------------------------------------------------------- Single NAL Unit Mode yes no Non-Interleaved Mode yes yes Interleaved Mode no no Table 3 Summary of allowed session-specific packetization modes for the I-C packetization mode (yes = allowed, no = disallowed) Session-Specific Mode Base Session Enhancement Session ---------------------------------------------------------- Single NAL Unit Mode no no Non-Interleaved Mode no no Interleaved Mode yes yes Table 4 summarizes the allowed NAL unit types for each allowed session-specific packetization mode of the NI-T packetization mode. Table 5 summarizes the allowed NAL unit types for each allowed session-specific packetization mode of the NI-C and NI-TC packetization modes. Table 6 summarizes the allowed NAL unit types Wenger, et al Expires December 30, 2008 [Page 23] Internet-Draft RTP Payload Format for SVC Video June 2008 for the only allowed session-specific packetization mode (i.e. the interleaved mode) of the I-C packetization mode. Table 4 Summary of allowed NAL unit types for each session-specific packetization mode of the NI-T packetization mode (yes = allowed, no = disallowed, ig = ignore) Type Packet Single NAL Non-Interleaved Unit Mode Mode ------------------------------------------------ 0 undefined ig ig 1-23 NAL unit yes yes 24 STAP-A no yes 25 STAP-B no no 26 MTAP16 no no 27 MTAP24 no no 28 FU-A no yes 29 FU-B no no 30 PACSI no yes 31 NI-MTAP no yes Table 5 Summary of allowed NAL unit types for each session-specific packetization mode of the NI-C and NI-TC packetization modes (yes = allowed, no = disallowed, ig = ignore) Type Packet Single NAL Non-Interleaved Unit Mode Mode ------------------------------------------------ 0 undefined ig ig 1-23 NAL unit yes yes 24 STAP-A no yes 25 STAP-B no no 26 MTAP16 no no 27 MTAP24 no no 28 FU-A no yes 29 FU-B no no 30 PACSI yes yes 31 NI-MTAP no yes Wenger, et al Expires December 30, 2008 [Page 24] Internet-Draft RTP Payload Format for SVC Video June 2008 Table 6 Summary of allowed NAL unit types for the session-specific packetization mode of the I-C packetization mode (yes = allowed, no = disallowed, ig = ignore) Type Packet Interleaved Mode ------------------------------ 0 undefined ig 1-23 NAL unit no 24 STAP-A no 25 STAP-B yes 26 MTAP16 yes 27 MTAP24 yes 28 FU-A yes 29 FU-B yes 30 PACSI yes 31 undefined no The NAL unit type values indicated as undefined in Tables 3.3, 3.4 and 3.5 are reserved for future extensions. NAL units of those types SHOULD NOT be sent by a sender and MUST be ignored by a receiver. Note that NAL unit type 30 and 31 are indicated as undefined in RFC 3984, therefore RFC 3984 receivers MUST ignore NAL units of this type, if present. 6.6 Aggregation Packets Please see Section 5.7 of [RFC3984]. 6.7 Fragmentation Units (FUs) Please see section 5.8 of [RFC3984]. 6.8 Payload Content Scalability Information (PACSI) NAL Unit One of the two new NAL unit types specified in this memo is the Payload Content Scalability Information (PACSI) NAL unit. The OPTIONAL PACSI NAL unit, if present, MUST be the first NAL unit in an aggregation packet or the NAL unit in a single NAL unit packet, and it MUST NOT be present in other types of packets. The PACSI NAL unit, when included in an aggregation packet, indicates scalability information and other characteristics that are common for all the remaining NAL units in the payload of the aggregation packet. Furthermore, a PACSI NAL unit MAY contain zero or more SEI NAL units. The PACSI NAL unit makes it easier for MANEs to decide Wenger, et al Expires December 30, 2008 [Page 25] Internet-Draft RTP Payload Format for SVC Video June 2008 whether to forward/process/discard the aggregation packet containing the PACSI NAL unit. Additional reasons to use PACSI NAL units are indicated later on, in the specification of the semantics of the fields. Senders MAY create PACSI NAL units and receivers MAY ignore them, or use them as hints to enable efficient aggregation packet processing or decoding order recovery in multi-source transmission. Note that the NAL unit type for the PACSI NAL unit (type 30) is among the types that are left unspecified in [H.264] and [RFC3984]. When the first aggregation unit of an aggregation packet contains a PACSI NAL unit, there MUST be at least one additional aggregation unit present in the same packet. The RTP header and payload header fields of the aggregation packet are set according to the remaining NAL units in the aggregation packet. When a PACSI NAL unit is included in a multi-time aggregation packet (MTAP), the decoding order number (DON) for the PACSI NAL unit MUST be set to indicate that the PACSI NAL unit has an identical DON to the first NAL unit in decoding order among the remaining NAL units in the aggregation packet. When a PACSI NAL unit is included in a single NAL unit packet, the RTP header and payload header fields of the packet are set according to the next non-PACSI NAL unit in transmission order. The structure of a PACSI NAL unit is as follows. The first four octets are exactly the same as the four-byte SVC NAL unit header discussed in Section 5.35.3. They are followed by one octet containing several flags, then five optional octets, and finally zero or more SEI NAL units. Each SEI NAL unit is preceded by a 16- bit unsigned size field (in network byte order) that indicates the size of the following NAL unit in bytes (excluding these two octets, but including the NAL unit type octet of the SEI NAL unit). Figure 1 illustrates the PACSI NAL unit structure and an example of a PACSI NAL unit containing two SEI NAL units. The bits A, P, C, S, and E are specified only if the bit X is equal to 1. The fields TL0PICIDX and IDRPICID are present only if the bit Y is equal to 1. The field DONC is present only if the bit T is equal to 1. The field T MUST be equal to 0 if the aggregation packet containing the PACSI NAL unit is not an STAP-A packet. Wenger, et al Expires December 30, 2008 [Page 26] Internet-Draft RTP Payload Format for SVC Video June 2008 0 1 2 3 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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |F|NRI| Type |R|I| PRID |N| DID | QID | TID |U|D|O| RR| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |X|Y|T|A|P|C|S|E| TL0PICIDX (o.)| IDRPICID (o.) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | DONC (o.) | NAL unit size 1 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | | SEI NAL unit 1 | | | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | NAL unit size 2 | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | | | SEI NAL unit 2 | | +-+-+-+-+-+-+-+-+-+-+ | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Figure 1 PACSI NAL unit structure. Fields suffixed by "(o.)" are OPTIONAL. The values of the fields in PACSI NAL unit MUST be set as follows. The term "target NAL unit" is used in the semantics of some fields. When the PACSI NAL unit is included in an aggregation packet, a "target NAL unit" refers to one or more NAL units that are contained in the aggregation packet, but not included in the PACSI NAL unit itself, that are in the same access unit as the first NAL unit following the PACSI NAL unit in the aggregation packet. When the PACSI NAL unit is included in a single NAL unit packet, a "target NAL unit" refers to the next non-PACSI NAL unit in transmission order. o The F bit MUST be set to 1 if the F bit in at least one of the remaining NAL units in the payload of the aggregation packet is equal to 1 (when the PACSI NAL unit is included in an aggregation packet) or if the next non-PACSI NAL unit in transmission order has the F bit equal to 1 (when the PACSI NAL unit is included in a single NAL unit packet). Otherwise, the F bit MUST be set to 0. Wenger, et al Expires December 30, 2008 [Page 27] Internet-Draft RTP Payload Format for SVC Video June 2008 o The NRI field MUST be set to the highest value of NRI field among all the remaining NAL units in the payload of the aggregation packet (when the PACSI NAL unit is included in an aggregation packet) or the value of the NRI field of the next non-PACSI NAL unit in transmission order (when the PACSI NAL unit is included in a single NAL unit packet). o The Type field MUST be set to 30. o The R bit MUST be set to 1. Receivers SHOULD ignore the value of R. o The I bit MUST be set to 1 if the I bit of at least one of the remaining NAL units in the payload of the aggregation packet is equal to 1 (when the PACSI NAL unit is included in an aggregation packet) or if the I bit of the next non-PACSI NAL unit in transmission order is equal to 1 (when the PACSI NAL unit is included in a single NAL unit packet). Otherwise, the I bit MUST be set to 0. o The PRID field MUST be set to the lowest value of the PRID values of all the remaining NAL units in the payload of the aggregation packet (when the PACSI NAL unit is included in an aggregation packet) or the PRID value of the next non-PACSI NAL unit in transmission order (when the PACSI NAL unit is included in a single NAL unit packet). o The N bit MUST be set to 1 if the N bit of all the remaining NAL units in the payload of the aggregation packet is equal to 1 (when the PACSI NAL unit is included in an aggregation packet) or if the N bit of the next non-PACSI NAL unit in transmission order is equal to 1 (when the PACSI NAL unit is included in a single NAL unit packet). Otherwise, the N bit MUST be set to 0. o The DID field MUST be set to the lowest value of the DID values of all the remaining NAL units in the payload of the aggregation packet (when the PACSI NAL unit is included in an aggregation packet) or the DID value of the next non-PACSI NAL unit in transmission order (when the PACSI NAL unit is included in a single NAL unit packet). o The QID field MUST be set to the lowest value of the QID values of all the remaining NAL units with the lowest value of DID in the payload of the aggregation packet (when the PACSI NAL unit is included in an aggregation packet) or the QID value of the next non-PACSI NAL unit in transmission order (when the PACSI NAL unit is included in a single NAL unit packet). Wenger, et al Expires December 30, 2008 [Page 28] Internet-Draft RTP Payload Format for SVC Video June 2008 o The TID field MUST be set to the lowest value of the TID values of all the remaining NAL units with the lowest value of DID in the payload of the aggregation packet (when the PACSI NAL unit is included in an aggregation packet) or the TID value of the next non-PACSI NAL unit in transmission order (when the PACSI NAL unit is included in a single NAL unit packet). o The U bit MUST be set to 1 if the U bit of at least one of the remaining NAL units in the payload of the aggregation packet is equal to 1 (when the PACSI NAL unit is included in an aggregation packet) or if the U bit of the next non-PACSI NAL unit in transmission order is equal to 1 (when the PACSI NAL unit is included in a single NAL unit packet). Otherwise, the U bit MUST be set to 0. o The D bit MUST be set to 1 if the D value of all the remaining NAL unit in the payload is equal to 1 (when the PACSI NAL unit is included in an aggregation packet) or if the D bit of the next non-PACSI NAL unit in transmission order is equal to 1 (when the PACSI NAL unit is included in a single NAL unit packet). Otherwise, the D bit MUST be set to 0. o The O bit MUST be set to 1 if the O bit of at least one of the remaining NAL units in the payload of the aggregation packet is equal to 1 (when the PACSI NAL unit is included in an aggregation packet) or if the O bit of the next non-PACSI NAL unit in transmission order is equal to 1 (when the PACSI NAL unit is included in a single NAL unit packet). Otherwise, the O bit MUST be set to 0. o The RR field MUST be set to '11' (in binary form). Receivers SHOULD ignore the value of RR. o If the X bit is equal to 1, the bits A, P, and C are specified as below. Otherwise, the bits A, P, and C are unspecified, and receivers MUST ignore these bits. The X bit SHOULD be identical for all the PACSI NAL units in all the RTP sessions carrying the same SVC bitstream. o If the Y bit is equal to 1, the OPTIONAL fields TL0PICIDX and IDRPICID MUST be present and specified as below, and the bits S and E are also specified as below. Otherwise, the fields TL0PICIDX and IDRPICID MUST NOT be present, whereas the S and E bits are unspecified and receivers MUST ignore these bits. The Y bit SHOULD be identical for all the PACSI NAL units in all the RTP sessions carrying the same SVC bitstream. Wenger, et al Expires December 30, 2008 [Page 29] Internet-Draft RTP Payload Format for SVC Video June 2008 o If the T bit is equal to 1, the OPTIONAL field DONC MUST be present and specified as below. Otherwise, the field DONC MUST NOT be present. o The A bit MUST be set to 1 if all the target NAL units belong to anchor layer representations. Otherwise, the A bit MUST be set to 0. The A bit SHOULD be identical for all the PACSI NAL units for which the target NAL units belong to the same access unit. Informative note: The A bit indicates whether CGS or spatial layer switching at a non-IDR layer representation (a layer representation with nal_unit_type not equal to 5 and idr_flag not equal to 1) can be performed. When a picture coding structure such as IBBP is in use, a non-IDR intra layer representation can be used for random access. Compared to using only IDR layer representations, higher coding efficiency can be achieved. The H.264/AVC or SVC solution to indicate the random accessibility of a non-IDR intra layer representation is using a recovery point SEI message. The A bit offers direct access to this information, without having to parse the recovery point SEI message, which may be buried deeply in an SEI NAL unit. Furthermore, the SEI message may not be present in the bitstream. o The P bit MUST be set to 1 if all the remaining NAL units in the payload of the aggregation packet have redundant_pic_cnt greater than 0 (when the PACSI NAL unit is included in an aggregation packet) or the next non-PACSI NAL unit in transmission order has redundant_pic_cnt greater than 0 (when the PACSI NAL unit is included in a single NAL unit packet). Otherwise, the P bit MUST be set to 0. Informative note: The P bit indicates whether a packet can be discarded because it contains only redundant slice NAL units. Without this bit, the corresponding information can be obtained from the syntax element redundant_pic_cnt, which is containedin the variable-length coded slice header. o The C bit MUST be set to 1 if all the target NAL units belong to intra layer representations. Otherwise, the C bit MUST be set to 0. The C bit SHOULD be identical for all the PACSI NAL units for which the target NAL units belong to the same access unit. Informative note: The C bit indicates whether a packet contains intra slices, which may be the only packets to be forwarded,, e.g. when the network conditions are particularly adverse. Wenger, et al Expires December 30, 2008 [Page 30] Internet-Draft RTP Payload Format for SVC Video June 2008 o The S bit MUST be set to 1, if the first VCL NAL unit, in decoding order, of the layer representation containing the first NAL unit following the PACSI NAL unit in the aggregation packet is present in the payload (when the PACSI NAL unit is included in an aggregation packet) or if the next non-PACSI NAL unit in transmission order is the first VCL NAL unit, in decoding order, of a layer representation (when the PACSI NAL unit is included in a single NAL unit packet). Otherwise, the S bit MUST be set to 0. o The E bit MUST be set to 1, if the last VCL NAL unit, in decoding order, of the layer representation containing the first NAL unit following the PACSI NAL unit in the aggregation packet is present in the payload (when the PACSI NAL unit is included in an aggregation packet) or if the next non-PACSI NAL unit in transmission order is the last VCL NAL unit, in decoding order, of a layer representation (when the PACSI NAL unit is included in a single NAL unit packet). Otherwise, the E field MUST be set to 0. Informative note: The S or E bit indicates whether the first or last slice, in transmission order, of a layer representation is in a packet, to enable a MANE to detect slice loss and take proper action such as requesting a retransmission as soon as possible, as well as to allow efficient playout buffer handling similarly to the M bit present in the RTP header. The M bit in the RTP header still indicates the end of an access unit, not the end of a layer representation. o When present, the TL0PICIDX field MUST be set to equal to tl0_dep_rep_idx as specified in Annex G of [H.264] for the layer representation containing the first NAL unit following the PACSI NAL unit in the aggregation packet (when the PACSI NAL unit is included in an aggregation packet) or containing the next non- PACSI NAL unit in transmission order (when the PACSI NAL unit is included in a single NAL unit packet). o When present, the IDRPICID field MUST be set to equal to effective_idr_pic_id as specified in Annex G of [H.264] for the layer representation containing the first NAL unit following the PACSI NAL unit in the aggregation packet (when the PACSI NAL unit is included in an aggregation packet) or containing the next non- PACSI NAL unit in transmission order (when the PACSI NAL unit is included in a single NAL unit packet). Informative note: The TL0PICIDX and IDRPICID fields enable the detection of the loss of layer representations in the most Wenger, et al Expires December 30, 2008 [Page 31] Internet-Draft RTP Payload Format for SVC Video June 2008 important temporal layer (0) by receivers as well as MANEs. SVC provides a solution that uses SEI messages, which are harder to parse and may not be present in the bitstream at all. o When present, the field DONC indicates the Cross-Session Decoding Order Number (CS-DON) for the first NAL unit of the remaining NAL units in the aggregation packet (when the PACSI NAL unit is included in an aggregation packet) or the CS-DON of the next non- PACSI NAL unit in transmission order (when the PACSI NAL unit is included in a single NAL unit packet). The CS-DON is further discussed in Section 6.10. The PACSI NAL unit SHALL include a subset (zero to all) of the SEI NAL units associated with the access unit to which the target NAL units belong, and SHALL NOT contain SEI NAL units associated with any other access unit. [Ed. (AE): Is the intention here to say: if the AU has SEI messages, then they must all be included in the PACSI. Or to say that the PACSI MAY include one or more of the SEI NAL units..., i.e., to make it an option? The Informative note below seems to indicate the latter (it uses the word "may").] Informative note: Senders may repeat such SEI NAL units in the PACSI NAL unit, so that they are repeated in more than one packet and thus increase robustness against packet loss. Receivers may use the repeated SEI messages in place of missing SEI messages. In H.264/AVC and SVC, within each access unit, SEI NAL units must appear before any VCL NAL unit in decoding order. Therefore, without using PACSI NAL units, SEI messages are typically only conveyed in the first of the packets comprising an access unit. When the PACSI NAL unit is included in an aggregation packet, an SEI message SHOULD NOT be included in the PACSI NAL unit and included in one of the remaining NAL units contained in the same aggregation packet. 6.9 Non-Interleaved Multi-Time Aggregation Packets (NI-MTAPs) The second new NAL unit type introduced in this memois the Non- Interleaved Multi-Time Aggregation packet (NI-MTAP). An NI-MTAP consists of zero or more non-interleaved multi-time aggregation units, as shown in Figure 2. Informative note: The rule above differs from the constraint on aggregation packets present in [RFC3984], where at least one NAL unit must be contained in the aggregation packet. Wenger, et al Expires December 30, 2008 [Page 32] Internet-Draft RTP Payload Format for SVC Video June 2008 The NI-MTAP consists of 16 bits of unsigned size information of the following NAL unit (in network byte order), and 16 bits (in network byte order) of timestamp offset (TS offset) for this NAL unit. The structure of the multi-time aggregation units for the NI-MTAP is presented in Figure 2. The starting or ending position of an aggregation unit within a packet MAY not be on a 32-bit word boundary. The NAL units in the NI-MTAP are ordered in NAL unit decoding order. The term NAL unit Effective Timestamp (ETS) is defined as the value that the RTP timestamp would have if the particular NAL unit was transported in its own RTP packet. This value is different from the actual RTP timestamp present in the packet carrying the particular NAL units in MTAP packets. Let ETS be the effective timestamp of a NAL unit and TS the actual RTP timestamp of the packet carrying the NAL unit. The timestamp offset field MUST be set to a value equal to the value of the following formula: If ETS >= TS, then TS offset = ETS - TS. If ETS < TS, then TS offset = ETS + (2^32 - TS). 0 1 2 3 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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ : NAL unit size | TS offset | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | | NAL unit | | | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | : +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Figure 2 Non-interleaved Multi-time aggregation unit for NI-MTAP For the "earliest" multi-time aggregation unit in an MTAP the timestamp offset MUST be zero. Hence, the RTP timestamp of the MTAP itself is identical to the earliest NAL unit effective timestamp. Informative note: The "earliest" multi-time aggregation unit is the one that would have the smallest extended RTP timestamp among all the aggregation units of an MTAP if the aggregation units were encapsulated in single NAL unit packets. An extended timestamp is a timestamp that has more than 32 bits and is capable of counting the wraparound of the timestamp field, thus enabling one to determine Wenger, et al Expires December 30, 2008 [Page 33] Internet-Draft RTP Payload Format for SVC Video June 2008 the smallest value if the timestamp wraps. Such an "earliest" aggregation unit may not be the first one in the order in which the aggregation units are encapsulated in an NI-MTAP. The "earliest" NAL unit need not be the same as the first NAL unit in the NAL unit decoding order either. Figure 3 presents an example of an RTP packet that contains an NI- MTAP multi-time aggregation packet that contains two non-interleaved multi-time aggregation units, labeled as 1 and 2 in the figure. 0 1 2 3 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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | RTP Header | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |NI-MTAPNAL HDR | NALU 1 Size | NALU 1 TS Off.| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | NALU 1 TS Off.| NALU 1 HDR | NALU 1 DATA | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ + : : ++-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-++ | NALU 2 SIZE | NALU 2 TS Offset | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | NALU 2 HDR | NALU 2 DATA | +-+-+-+-+-+-+-+-+ | : : | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | :...OPTIONAL RTP padding | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Figure 3 An RTP packet including a NI-MTAP non-interleaved multi-time aggregation packet and two non-interleaved multi- time aggregation units 6.10 Decoding Order Number (DON) The DON concept is introduced in RFC 3984 and is used to recover the decoding order when interleaving is used within a single session. Section 5.5 of [RFC3984] applies when using SST. When using MST, it is necessary to recover the decoding order across the various RTP sessions regardless if interleaving is used or not. In addition to the timestamp mechanism desribed later on, the CS-DON mechanism is an extension of the DON facility that can be used for this purpose, and is defined in the following section. Wenger, et al Expires December 30, 2008 [Page 34] Internet-Draft RTP Payload Format for SVC Video June 2008 6.10.1 Cross-Session DON (CS-DON) for Multi-Source Transmission The Cross-Session Decoding Order Number (CS-DON) is a number that indicates the decoding order of NAL units across all sessions involved in MST. It is similar to the DON concept in [H.264], but contrary to RFC 3984 where the DON was used only for interleaved packetization, in this memo it is used not only in the interleaved mode (I-C) but also in two of the non-interleaved modes as well (NI- C and NI-TC). When the NI-C or NI-TC MST packetization modes are in use, the packetization of each session MUST be as specified in Section 7.1. In PACSI NAL units the CS-DON value is explicitly coded in the field DONC. For non-PACSI NAL units the CS-DON value is derived as follows. Let SN indicate the RTP sequence number of a packet, and recall that the NAL unit effective timestamp (ETS) was defined in Section 6.9 as the value that the RTP timestamp would have if that NAL unit would be transported in its own RTP packet. o For each non-PACSI NAL unit carried in a session using the single NAL unit session-specific packetization mode, the CS-DON value of the NAL unit is equal to (DONC_prev_PACSI + SN_diff - 1) % 65536, wherein '%' is the modulo operation, DONC_prev_PACSI is the DONC value of the previous PACSI NAL unit with the same ETS as the current NAL unit, and SN_diff is calculated as follows: if SN1 > SN2, SN_diff = SN1 - SN2 else SN_diff = SN2 + 65536 - SN1 where SN1 and SN2 are the SNs of the current NAL unit and the previous PACSI NAL unit with the same ETS, respectively. o For non-PACSI NAL units carried in a session using a non- interleaved session-specific packetization mode (NI-TC, NI-C), the CS-DON value of each non-PACSI NAL unit is derived as follows. . For a non-PACSI NAL unit in a single NAL unit packet, the following applies. . If the previous PACSI NAL unit is contained in a single NAL unit packet, the CS-DON value of the NAL unit is calculated as above when the single NAL unit session- specific packetization mode is in use; Wenger, et al Expires December 30, 2008 [Page 35] Internet-Draft RTP Payload Format for SVC Video June 2008 . otherwise (the previous PACSI NAL unit is contained in an STAP-A packet), the CS-DON value of the NAL unit is equal to: (the CS-DON value of the previous non-PACSI NAL unit in decoding order + 1) % 65536, where '%' is the modulo operation. . For a non-PACSI NAL unit in an STAP-A packet, the following applies. . If the non-PACSI NAL unit is the first non-PACSI NAL unit in the STAP-A packet, the CS-DON value of the NAL unit is equal to DONC of the PACSI NAL unit in the STAP- A packet; . otherwise (the non-PACSI NAL unit is not the first non- PACSI NAL unit in the STAP-A packet), the CS-DON value of the NAL unit is equal to: (the CS-DON value of the previous non-PACSI NAL unit in decoding order + 1) % 65536, wherein '%' is the modulo operation. . For a non-PACSI NAL unit in a number of FU-A packets, the CS- DON value of the NAL unit is calculated as above when the single NAL unit session-specific packetization mode is in use, with SN1 being the SN value of the first FU-A packet. When the I-C MST packetization mode is in use, the DON values derived according to RFC 3984 of all the NAL units in each of the multiplexed RTP sessions MUST indicate CS-DON values. 7. Packetization Rules Section 6 of [RFC3984] applies in this memo, with the following additions. All receivers MUST support the single NAL unit packetization mode to provide backward compatibility to endpoints supporting only the single NAL unit mode of RFC 3984. However, the use of single NAL unit packetization mode (packetization-mode equal to 0) SHOULD be avoided whenever possible, because encapsulating NAL units of small sizes in their own packets (e.g. small NAL units containing parameter sets, prefix NAL units, or SEI messages) is less efficient due to the packet header overhead. All receivers MUST support the non-interleaved mode of [RFC3984]. Informative note: The non-interleaved mode does allow an application to encapsulate a single NAL unit in a single RTP Wenger, et al Expires December 30, 2008 [Page 36] Internet-Draft RTP Payload Format for SVC Video June 2008 packet. Historically, the single NAL unit mode has been included into [RFC3984] only for compatibility with ITU-T Rec. H.241 Annex A [H.241]. There is no point in carrying this historic ballast towards a new application space such as the one provided with SVC. The implementation complexity increase for supporting the additional mechanisms of the non-interleaved mode (namely STAP-A and FU-A) is minor, whereas the benefits are significant. As a result, STAP-A and FU-A implementation is required. A NAL unit of small size SHOULD be encapsulated in an aggregation packet together with one or more other NAL units. For example, non- VCL NAL units such as access unit delimiter, parameter set, or SEI NAL unit are typically small. A prefix NAL unit and the NAL unit with which it is associated, and which follows the prefix NAL unit in decoding order, SHOULD be included in the same aggregation packet whenever an aggregation packet is used for the associated NAL unit. Informative note: Although the prefix NAL unit is ignored by an H.264/AVC decoder, it is necessary in the SVC decoding process. Given the small size of the prefix NAL unit, it is best if it is transported in the same RTP packet as its associated NAL unit. When the first aggregation unit of an aggregation packet contains a PACSI NAL unit, there MUST be at least one additional aggregation unit present in the same packet. 7.1 Packetization Rules for Multi-Source Transmission When MST is used, decoding order recovery for NAL units carried in the associated RTP sessions is needed. The following packetization rules ensure that decoding order of NAL units carried in the sessions can be correctly recovered for each of the MST packetization modes using the de-packetization process specified in Section 8.1. The NI-T and NI-TC modes rely on timestamps to recover the decoding order. In order to be able to do so, it is necessary for the SVC stream to contain data for all sampling instances of a given layer in all enhancement layers that depend on the given layer. The NI-TC, NI-C, and I-C modes do not have this limitation, and use the CS-DON value as a means to explicitly indicate decoding order, either direcly coded in PACSI NAL units, or inferred from them using the packetization rules. It is noted that the NI-TC mode offers both techniques and it is up to the receiver to select which one to use. Wenger, et al Expires December 30, 2008 [Page 37] Internet-Draft RTP Payload Format for SVC Video June 2008 7.1.1 NI-T / NI-TC Packetization Rules When the NI-T or NI-TC packetization mode is in use, the following applies. o If one or more NAL units of an access unit of sampling time instance t is present in RTP session A, then one or more NAL units of the same access unit of the same sampling time instance t MUST be present in any enhancement RTP session which depends on RTP session A. Informative note: There are multiple ways to insert additional NAL units in order to satisfy this rule: - One option for adding additional NAL units is to place NI- MTAP packets (defined in Section 6.9), and not include any aggregation packet in the payload. Although empty, these packets are used by the process described in Section 8.1.1 for the access unit re-ordering process. - Additional NAL units may also be added by repeating prefix NAL units (NAL unit type 14). Before passing NAL units to the decoder re-ordering of the access unit as described in Section 8.1.1 is needed. This may only be possible for access units which contain base layer NAL units. [Ed. (TS): It may be useful to indicate in the SDP parameters that additional NAL unit re-ordering as specified in 7.1.4 is not required.][Ed. (AE): I don't understand this comment.] - Additional NAL units may also be added by placing single NAL unit packets containing exactly one PACSI NAL unit in the enhancement RTP sessions. - Additional NAL units may also be added by the encoder itself. This option, however, may not be available with pre- encoded content. o When not using NI-TC mode and a PACSI NAL unit is present, the T bit MUST be equal to 0, i.e. the DONC field MUST NOT be present. 7.1.2 NI-C / NI-TC Packetization Rules When the NI-C or NI-TC packetization mode is in use, the following applies. Wenger, et al Expires December 30, 2008 [Page 38] Internet-Draft RTP Payload Format for SVC Video June 2008 o For each single NAL unit packet containing a non-PACSI NAL unit, the previous packet, if present, MUST have the same RTP timestamp as the single NAL unit packet, and the following applies. . If the ETS of the non-PACSI NAL unit is not equal to the ETS of the previous non-PACSI NAL unit in decoding order, the previous packet MUST contain a PACSI NAL unit containing the DONC field; . Otherwise (the ETS of the non-PACSI NAL unit is equal to the ETS of the previous non-PACSI NAL unit in decoding order), the previous packet MAY contain a PACSI NAL unit containing the DONC field. o For each STAP-A packet, the first NAL unit in the STAP-A packet, if present, MUST be a PACSI NAL unit containing the DONC field. [Ed. (AE): Is it possible to have an empty STAP-A? Or was the "if present" superfluous?] o For each FU-A packet, if present, the previous packet MUST have the same RTP timestamp as the FU-A packet, and the following applies. [Ed. (AE): See the previous comment for STAP-A, regarding the "if present" part.] . If the FU-A packet is the start of the fragmented NAL unit, the following applies; . If the ETS of the fragmented NAL unit is not equal to the ETS of the previous non-PACSI NAL unit in decoding order, the previous packet MUST contain a PACSI NAL unit containing the DONC field; . Otherwise (the ETS of the fragmented NAL unit is equal to the ETS of the previous non-PACSI NAL unit in decoding order), the previous packet MAY contain a PACSI NAL unit containing the DONC field. . Otherwise if the FU-A packet is the end of the fragmented NAL unit, the following applies. . If the next non-PACSI NAL unit in decoding order has ETS equal to the ETS of the fragmented NAL unit, and is carried in a number of FU-A packets or a single NAL unit packet, the next packet MUST be a single NAL unit packet containing a PACSI NAL unit containing the DONC field. [Ed. (AE): Does this mean I am inserting a single RTP packet with just PACSI in it? Just to make sure.] Wenger, et al Expires December 30, 2008 [Page 39] Internet-Draft RTP Payload Format for SVC Video June 2008 . Otherwise (the FU-A packet is neither the start nor the end of the fragmented NAL unit), the previous packet MUST be a FU-A packet. o For each single NAL unit packet containing a PACSI NAL unit, if present, the PACSI NAL unit MUST contain the DONC field. 7.1.3 I-C Packetization Rules When the I-C session-multiplexing packetization mode is in use, the following applies. o When a PACSI NAL unit is present, the T bit MUST be equal to 0, i.e., the DONC field MUST NOT be present.[Ed. (AE): Why? Revisit.] 7.1.4 Packetization Rules for Non-VCL NAL Units NAL units which do not directly encode video slices are known in H.264 as non-VCL NAL units. Non-VCL units that are only used by, or only relevant to, enhancement RTP sessions SHOULD be sent in the lowest session to which they are relevant. Some senders, however, such as those sending pre-encoded data, might not be able to easily determine which non-VCL units are relevant to which session. Thus, essential non-VCL NAL units (parameter sets sent in-band, i.e., NAL unit types 7, 8, 13, and 15) MAY, instead, be sent in session that the one they are used by depends on (e.g., the base RTP session), and non-essential non-VCL NAL units MAY be sent in any RTP session. If a non-VCL unit is relevant to more than one RTP session, neither of which depends on the other(s), the NAL unit MAY be sent in another session which all these sessions depend on. Alternatively, it MAY be repeated in all such sessions. In general, identical non- VCL units MAY be sent in more than one session for redundancy. [Ed. (JL): Can this cause issues with HRD timing?] 7.1.5 Packetization Rules for Prefix NAL Units When the base layer is sent on an RTP session using the Non- Interleaved or the Interleaved mode, prefix NAL units SHOULD be aggregated (using STAP-A or STAP-B units) with the NAL unit they prefix, unless this would violate session MTU constraints or if fragmentation units are used. Wenger, et al Expires December 30, 2008 [Page 40] Internet-Draft RTP Payload Format for SVC Video June 2008 If the base layer is sent in a base RTP session using RFC 3984, prefix NAL units MAY be sent in the lowest enhancement RTP session rather than in the base RTP session. 8. De-Packetization Process For single-source transmission where a single RTP session is used, the de-packetization process specified in Section 7 of [RFC3984] applies. [Ed. (??): with some fixes to section 7 of RFC 3984 and some changes/additions to section 7.3 (Additional De-Packetization Guidelines) of RFC 3984 - TBD] For multi-source transmission, where more than one RTP sessions are used to receive data from the same SVC bitstream, the de- packetization process is specified in Section 8.1. 8.1 De-Packetization Process for Multi-Source Transmission As for a single RTP session, the general concept behind the de- packetization process is to reorder NAL units from transmission order to the NAL unit decoding order. The sessions to be received SHALL be identified by mechanisms specified in [I-D.ietf-mmusic-decoding-dependency]. Enhancement RTP sessions typically contain an RTP stream that depends on at least one other RTP session, as indicated by mechanisms defined in [I- D.ietf-mmusic-decoding-dependency]. A lower RTP session to an enhancement RTP session is an RTP session which the enhancement RTP session depends on. The lowest RTP session for a receiver is the base RTP session, which does not depend on any other RTP session received by the receiver. The highest RTP session for a receiver is the RTP session which no other RTP session received by the receiver depends on. For each of the RTP sessions, the RTP reception process as specified in RFC 3550 is applied. Then the received packets are passed in increasing order of sequence number into the payload de- packetization to NAL units as defined in this memo. The decoding order of the NAL units carried in all the associated RTP sessions is then recovered by applying one of the following subsections, depending on which of the MST packetization modes is in use. Wenger, et al Expires December 30, 2008 [Page 41] Internet-Draft RTP Payload Format for SVC Video June 2008 8.1.1 Decoding Order Recovery for the NI-T and NI-TC Modes The following process SHALL be applied when the NI-T packetization mode is in use. The following process MAY be applied when the NI-TC packetization mode is in use. The process is based on RTP session dependency signaling, RTP sequence numbers, and timestamps. The decoding order of NAL units within an RTP packet stream in RTP session S is given by the ordering of sequence numbers SN of the RTP packets the NAL units are contained in. In an aggregation packet contained in an RTP packet the decoding order is given by the order of appearance of the NAL units within the packet. The RTP session identifier S gives the increasing order of dependency of the received RTP sessions as indicated by mechanisms specified in Section 9.2.3, where S equal to 0 identifies the base RTP session. [Ed. (AE): Does the mmusic draft excplicitly order the session ID's by dependency? I coulnd't find it in the text.] Timing information according to the media timestamp TS(SN) derived from the RTP packet timestamp of the RTP packet with sequence number SN is associated with all NAL units contained in the same RTP packet received in RTP session S. For NI-MTAP packets the effective timstamp ETS is derived for each contained NAL unit by using the "TS offset" value in the NI-MTAP packet as defined in 6.9, and is used instead of the actual TS. NAL units contained in fragmentation packets are handled as defragmented, entire NAL units with their own timestamp. All NAL units associated with the same value of media timestamp TS are part of the same access unit AU(TS). Each NI-MTAP packet which does not contain any aggregation units or each PACSI NAL unit in a single NAL unit packet SHOULD be kept as, effectively, access unit indicators in the re-ordering process. NI- MTAP or PACSI NAL units SHOULD be removed before passing access unit data to the decoder. Informative Note: These special (essentially, empty) NI-MTAP and PACSI NAL units are used to associate NAL units present in other RTP sessions with RTP sessions not containing any data for an access unit of a particular time instance. They act as access unit indicators in sessions that would otherwise contain no data for the particular access unit. The presence of these NAL units is ensured by the packetization rules in Section 7.1.1. Wenger, et al Expires December 30, 2008 [Page 42] Internet-Draft RTP Payload Format for SVC Video June 2008 The decoding order of NAL units from multiple RTP streams in multiple RTP sessions SHALL be recovered into a single sequence of NAL units, grouped into access units, by performing the following steps: o The process SHOULD be started with the NAL units received in the highest RTP enhancement session with the earliest timestamp TS available in the session's (de-jittering) buffer. o Collect all NAL units associated with the same value of timestamp TS, starting from the highest RTP enhancement session, from all the (de-jittering) buffers of the received RTP sessions. The collected NAL units will be those associated with the access unit AU(TS). o Place the collected NAL units in increasing order of session identifier S. o Place the ordered collected NAL units in decoding order within the particular access unit by satisfying the NAL unit ordering rules for SVC access units, as specified in the informative algorithm provided in Section 8.1.1.1. o Remove NI-MTAP and any PACSI NAL units from the access unit AU(TS). o The access units MAY be transferred to the decoder. If access units AU(TS) are transferred to the decoder, they SHALL be passed in the order of appearance (given by the order of RTP sequence numbers) of timestamp values TS in the highest RTP session associated with access unit AU(TS). Informative Note: Due to packet loss it is possible that not all sessions may have NAL units present for the timestamp value TS present in the highest RTP session. In such a case an algorithm may: a) proceed to the next complete access unit with NAL units present in all the received RTP sessions; or b) consider the highest RTP session to be the largest session identifier for which the access unit is complete, and apply the process above. The algorithm may return to the original highest RTP session when a complete and error- free access unit that contains NAL units in all the sessions is received. Informative example: Wenger, et al Expires December 30, 2008 [Page 43] Internet-Draft RTP Payload Format for SVC Video June 2008 The example shown in Figure 4 refers to three RTP sessions A, B and C containing an SVC bitstream transmitted as 3 sources. In the example, the dependency signaling as described in Section 9.2.3, indicates that session A is the base RTP session, B is the first enhancement RTP session and depends on A, and C is the second RTP enhancement session and depends on A and B. A hierarchical picture coding prediction structure is used, in which Session A has the lowest frame rate and Session B and C have the same but higher frame rate. The figure indicates decoding order numbers for NAL units in the various packets across the sessions, as well as the associated media timestamps (TS[]). The example demonstrates decoding order recovery when differenta amounts of jitter is present in each of the sessions (i.e., i.e., at buffering startup not all packets with the same timestamp are available in all the de-jittering buffers. The process first proceeds to the NAL units associated with the first timestamp TS[1] present in session C and removes/ignores all preceding NAL units to NAL units with TS[1] in each of the de- jittering buffers of RTP sessions A, B, and C. Then, starting from session C, the first timestamp available in decoding order (TS [1]) is selected and NAL units starting from RTP session A, and sessions B and C are placed in order of the RTP session dependency (in the example for TS[1]: first session B and then session C) into the access unit AU(TS[1]) associated with timestamp TS[1]. Then the next timestamp TS[3] in order of appearance in the highest RTP session C is processed and the process described above is repeated. Note that there may be access units with no NAL units present, e.g., in the lowest RTP session A (see, e.g., TS[1]). With TS[8], the first access unit with NAL units present in all the RTP sessions appears in the buffers. C: ------------(1,2)-(3,4)--(5)---(6)---(7,8)(9,10)-(11)--(12)---- | | | | | | | | | | B: -(1,2)-(3,4)-(5)---(6)--(7,8)-(9,10)-(11)-(12)--(13,14)(15,15)- | | | | | | A: -------(1)---------------(2)---(3)---------------(4)----(5)---- -------------------------------------------------------------------> TS: [4] [2] [1] [3] [8] [6] [5] [7] [12] [10] Key: A, B, C - RTP sessions Integer values in '()' - NAL unit decoding order within RTP session '( )' - groups the NAL units of an access unit Wenger, et al Expires December 30, 2008 [Page 44] Internet-Draft RTP Payload Format for SVC Video June 2008 in a RTP session '|' - indicates corresponding NAL units of the same access unit AU(TS[..]) in the RTP sessions Integer values in '[]' - media Timestamp (TS), sampling time as derived from RTP timestamps associated to the access unit AU(TS[..]). Figure 4 Example of decoding order recovery in multi-source transmission with different session jitter at startup. 8.1.1.1 Informative Algorithm for NI-T Decoding Order Recovery within an Access Unit Within an access unit, the [H.264] specification (Sections 7.4.1.2.3 and G.7.4.1.2.3) constrains the valid decoding order of NAL units. These constraints make it possible to reconstruct a valid decoding order for the NAL units of an access unit based only on the order of NAL units in each session, the NAL unit headers, and Supplemental Enhancement Information message headers. This section specifies an informative algorithm to reconstruct a valid decoding order for NAL units within an access unit. Other NAL unit orderings may also be valid; however, any compliant NAL unit ordering will describe the same video stream and ancillary data as the one produced by this algorithm. An actual implementation, of course, needs only to behave "as if" this reordering is done. In particular, NAL units which are discarded by an implementation's decoding process do not need to be reordered. In this algorithm, NAL units within an access unit are first ordered by NAL unit type, in the order specified in the list below, except from NAL unit type 14 which is handled specially as described. NAL units of the same type are then ordered as specified for the type, if necessary. For the purposes of this algorithm, "session order" is the order of NAL units implied by their transmission order within an RTP session. For the Non-Interleaved and Single NAL unit modes, this is the RTP sequence number order coupled with the order of NAL units within an aggregation unit. o 9 Access unit delimiter Wenger, et al Expires December 30, 2008 [Page 45] Internet-Draft RTP Payload Format for SVC Video June 2008 Only one access unit delimiter will be present within an access unit. o 7 Sequence parameter set Any order of sequence parameter sets within an access unit is valid. o 13 Sequence parameter set extension Any order of sequence parameter set extensions within an access unit is valid. o 15 Subset sequence parameter set Any order of subset sequence parameter sets within an access unit is valid. o 8 Picture parameter set Any order of picture parameter sets within an access unit is valid. o 6 Supplemental enhancement information (SEI) If an SEI message with a first payload of 0 (Buffering Period) is present, it must be the first SEI. If SEI messages with a Scalable Nesting (30) payload and a nested payload of 0 (Buffering Period) are present, these then follow. Such an SEI message with the all_layer_representations_in_au_flag equal to 1 is placed first, followed by any others, sorted in DQId order by the highest DQId mentioned. All other SEI messages follow in any order. o 1 Coded slice of a non-IDR picture o 5 Coded slice of an IDR picture o 14 Prefix NAL unit in scalable extension NAL units of type 1 or 5 will be sent within only a single session for any given access unit. They are placed in session order. (Note: Any given access unit will contain only NAL units of type 1 or type 5, not both.) Wenger, et al Expires December 30, 2008 [Page 46] Internet-Draft RTP Payload Format for SVC Video June 2008 If NAL units of type 14 are present, every NAL unit of type 1 or 5 is prefixed by a NAL unit of type 14. (Note: Within an access unit, every NAL unit of type 14 is identical, so correlation of type 14 NAL units with the other NAL units is not necessary.) [Ed. (AE): Shouldn't this go before #5??]` o 12 Filler data o 14 Prefix NAL unit in scalable extension Any order of filler data units within an access unit is valid. [Ed. (AE): Shouldn't this move up, below #12?] If NAL units of type 14 are present, every filler data NAL unit is prefixed by a NAL unit of type 14. o 2 Coded slice data partition A o 3 Coded slice data partition B o 4 Coded slice data partition C These NAL units will be sent within only a single session for any given access unit, and are placed in session order. (Note: No current SVC profile uses slice data partitioning.) o 19 Coded slice of an auxiliary coded picture without partitioning These NAL units will be sent within only a single session for any given access unit, and are placed in session order. o 16-18 Reserved o 21-23 Reserved These are placed immediately following the non-reserved-type VCL NAL unit they follow in session order. o 20 Coded slice in scalable extension These are placed in DQId order, based on the dependency_id and quality_id values in the slice's NAL unit header extension. Within each DQId, they are placed in session order. (Note: Wenger, et al Expires December 30, 2008 [Page 47] Internet-Draft RTP Payload Format for SVC Video June 2008 SVC slices with a given DQId value will be sent on a single session for any given access unit.) o 10 End of sequence Only one end of sequence will be present within an access unit. o 11 End of stream Only one end of stream will be present within an access unit. [Ed. (AE): This list needs reformatting as a table.] 8.1.2 Decoding Order Recovery for the NI-C, NI-TC and I-C Modes The following process SHALL be used when either the NI-C or I-C MST packetization mode is in use. The following process MAY be applied when the NI-TC MST packetization mode is in use. The RTP packets output from the RTP-level reception processing for each session are placed into a remultiplexing buffer. [Ed.Note(YkW): Add handling of some cases of packet losses when the NI-C or NI-TC mode is in use, that discards some received NAL units for which the CS-DON value cannot be derived.] It is RECOMMENDED to set the size of the remultiplexing buffer (in bytes) equal to or greater than the value of the sprop-remux-buf-req media type parameter of the highest RTP session the receiver receives. The CS-DON value is calculated and stored for each NAL unit. The receiver operation is described below with the help of the following functions and constants: o Function AbsDON is specified in Section 8.1 of RFC 3984. o Function don_diff is specified in Section 5.5 of RFC 3984. o Constant N is the value of the OPTIONAL sprop-mst-interleave- depth media type parameter of the highest RTP session incremented by 1. Initial buffering lasts until one of the following conditions is fulfilled: Wenger, et al Expires December 30, 2008 [Page 48] Internet-Draft RTP Payload Format for SVC Video June 2008 o There are N or more VCL NAL units in the remultiplexing buffer. o If sprop-mst-max-don-diff of the highest RTP session is present, don_diff(m,n) is greater than the value of sprop-mst-max-don-diff of the highest RTP session, where n corresponds to the NAL unit having the greatest value of AbsDON among the received NAL units and m corresponds to the NAL unit having the smallest value of AbsDON among the received NAL units. o Initial buffering has lasted for the duration equal to or greater than the value of the OPTIONAL sprop-remux-init-buf-time media type parameter of the highest RTP session. The NAL units to be removed from the remultiplexing buffer are determined as follows: o If the remultiplexing buffer contains at least N VCL NAL units, NAL units are removed from the remultiplexing buffer and passed to the decoder in the order specified below until the buffer contains N-1 VCL NAL units. o If sprop-mst-max-don-diff of the highest RTP session is present, all NAL units m for which don_diff(m,n) is greater than sprop- max-don-diff of the highest RTP session are removed from the remultiplexing buffer and passed to the decoder in the order specified below. Herein, n corresponds to the NAL unit having the greatest value of AbsDON among the NAL units in the remultiplexing buffer. The order in which NAL units are passed to the decoder is specified as follows: o Let PDON be a variable that is initialized to 0 at the beginning of the RTP sessions. o For each NAL unit associated with a value of CS-DON, a CS-DON distance is calculated as follows. If the value of CS-DON of the NAL unit is larger than the value of PDON, the CS-DON distance is equal to CS-DON - PDON. Otherwise, the CS-DON distance is equal to 65535 - PDON + CS-DON + 1. o NAL units are delivered to the decoder in increasing order of CS- DON distance. If several NAL units share the same value of CS- DON distance, they can be passed to the decoder in any order. Wenger, et al Expires December 30, 2008 [Page 49] Internet-Draft RTP Payload Format for SVC Video June 2008 o When a desired number of NAL units have been passed to the decoder, the value of PDON is set to the value of CS-DON for the last NAL unit passed to the decoder. 9. Payload Format Parameters This section specifies the parameters that MAY be used to select optional features of the payload format and certain features of the bitstream. The parameters are specified here as part of the media type registration for the SVC codec. A mapping of the parameters into the Session Description Protocol (SDP) [RFC4566] is also provided for applications that use SDP. Equivalent parameters could be defined elsewhere for use with control protocols that do not use SDP. Some parameters provide a receiver with the properties of the stream that will be sent. The names of all these parameters start with "sprop" for stream properties. Some of these "sprop" parameters are limited by other payload or codec configuration parameters. For example, the sprop-parameter-sets parameter is constrained by the profile-level-id parameter. The media sender selects all "sprop" parameters rather than the receiver. This uncommon characteristic of the "sprop" parameters may not be compatible with some signaling protocol concepts, in which case the use of these parameters SHOULD be avoided. 9.1 Media Type Registration The media subtype for the SVC codec is allocated from the IETF tree. The receiver MUST ignore any unspecified parameter. Informative note: Requiring that the receiver ignores unspecified parameters allows for backward compatibility of future extensions. For example, if a future specification that is backward compatible to this specification specifies some new parameters, then a receiver according to this specification is capable of receiving data per the new payload but ignoring those parameters newly specified in the new payload specification. This provision is also present in RFC 3984. Media Type name: video Media subtype name: H264-SVC [Ed. (??): Text on "H264" must go into different section, see Colin's comments sent on 10 June 2008]The media subtype "H264" MUST Wenger, et al Expires December 30, 2008 [Page 50] Internet-Draft RTP Payload Format for SVC Video June 2008 be used for RTP streams using RFC 3984, i.e. not using any of the new features introduced by this specification compared to RFC 3984. [Edt. Note: The new features are to be listed herein.] For RTP streams using any of the new features introduced by this specification compared to RFC 3984, the media subtype "H264-SVC" SHOULD be used, and the media subtype "H264" MAY be used. Use of the media subtype "H264" for RTP streams using the new features allows for RFC 3984 receivers to negotiate and receive H.264/AVC or SVC streams packetized according to this specification, but to ignore media parameters and NAL unit types it does not recognize. Required parameters: none OPTIONAL parameters: profile-level-id: A base16 [RFC3548] (hexadecimal) representation of the following three bytes in the sequence parameter set NAL unit specified in [H.264]: 1) profile_idc, 2) a byte herein referred to as profile-iop, composed of the values of constraint_set0_flag, constraint_set1_flag, constraint_set2_flag, constraint_set3_flag, and reserved_zero_4bits in bit-significance order, starting from the most significant bit, and 3) level_idc. Note that reserved_zero_4bits is required to be equal to 0 in [H.264], but other values for it may be specified in the future by ITU- T or ISO/IEC. If the profile-level-id parameter is used to indicate properties of a NAL unit stream, it indicates the profile and level that a decoder has to support in order to comply with [H.264] when it decodes the NAL unit stream. The profile-iop byte indicates whether the NAL unit stream also obeys all the constraints as specified in subsection G.7.4.2.1.1 of [H.264]. Herein the NAL unit stream refers to the one consisting of all NAL units conveyed in the current RTP session, and all NAL units conveyed in other RTP sessions, if present, the current RTP session depends on. The current RTP session MAY depend on other RTP sessions when a scalable bitstream is transported with more than one RTP session and the current session is not an independent RTP session. If the profile-level-id parameter is used for capability exchange or session setup, it indicates the profile that the codec supports and the highest level supported for the signaled profile. The profile-iop byte indicates whether the codec has additional limitations whereby only the common Wenger, et al Expires December 30, 2008 [Page 51] Internet-Draft RTP Payload Format for SVC Video June 2008 subset of the algorithmic features and limitations signaled with the profile-iop byte is supported by the codec. For example, if a codec supports only the common subset of the coding tools of the Baseline profile and the Main profile at level 2.1 and below, the profile-level-id becomes 42E015, in which 42 stands for the Baseline profile, E0 indicates that only the common subset for all profiles is supported, and 15 indicates level 2.1. Informative note: Capability exchange and session setup procedures should provide means to list the capabilities for each supported codec profile separately. For example, the one-of-N codec selection procedure of the SDP Offer/Answer model can be used (section 10.2 of [RFC4566]). If no profile-level-id is present, the Baseline Profile without additional constraints at Level 1 MUST be implied. max-mbps, max-fs, max-cpb, max-dpb, and max-br: The common property of these parameters is as specified in RFC 3984. max-mbps: This parameter is as specified in RFC 3984. max-fs: This parameter is as specified in RFC 3984. max-cpb: The value of max-cpb is an integer indicating the maximum coded picture buffer size in units of 1000 bits for the VCL HRD parameters (see A.3.1 item i or G.10.2.2 item g of [H.264]) and in units of 1200 bits for the NAL HRD parameters (see A.3.1 item j or G.10.2.2 item h of [H.264]). The max-cpb parameter signals that the receiver has more memory than the minimum amount of coded picture buffer memory required by the signaled level conveyed in the value of the profile-level-id parameter. When max-cpb is signaled, the receiver MUST be able to decode NAL unit streams that conform to the signaled level, with the exception that the MaxCPB value in Table A-1 of [H.264] for the signaled level is replaced with the value of max-cpb. The value of max-cpb MUST be greater than or equal to the value of MaxCPB for the level given in Table A-1 of [H.264]. Senders MAY use this knowledge to construct coded video streams with greater variation of bit rate than can be achieved with the MaxCPB value in Table A-1 of [H.264]. Informative note: The coded picture buffer is used in the Hypothetical Reference Decoder (HRD, Annex C) of [H.264]. The use of the HRD is recommended in SVC encoders to verify Wenger, et al Expires December 30, 2008 [Page 52] Internet-Draft RTP Payload Format for SVC Video June 2008 that the produced bitstream conforms to the standard and to control the output bitrate. Thus, the coded picture buffer is conceptually independent of any other potential buffers in the receiver, including remultiplexing and de-jitter buffers. The coded picture buffer need not be implemented in decoders as specified in Annex C of [H.264]; standard- compliant decoders can have any buffering arrangements provided that they can decode standard-compliant bitstreams. Thus, in practice, the input buffer for video decoder can be integrated with the remultiplexing and de- jitter buffers of the receiver. max-dpb: This parameter is as specified in RFC 3984. max-br: The value of max-br is an integer indicating the maximum video bit rate in units of 1000 bits per second for the VCL HRD parameters (see A.3.1 item i or G.10.2.2 item g of [H.264]) and in units of 1200 bits per second for the NAL HRD parameters (see A.3.1 item j or G.10.2.2 item h of [H.264]). The max-br parameter signals that the video decoder of the receiver is capable of decoding video at a higher bit rate than is required by the signaled level conveyed in the value of the profile-level-id parameter. When max-br is signaled, the video codec of the receiver MUST be able to decode NAL unit streams that conform to the signaled level, conveyed in the profile-level-id parameter, with the following exceptions in the limits specified by the level: o The value of max-br replaces the MaxBR value of the signaled level (in Table A-1 of [H.264]). o When the max-cpb parameter is not present, the result of the following formula replaces the value of MaxCPB in Table A-1 of [H.264]: (MaxCPB of the signaled level) * max-br / (MaxBR of the signaled level). For example, if a receiver signals capability for Level 1.2 with max-br equal to 1550, this indicates a maximum video bitrate of 1550 kbits/sec for VCL HRD parameters, a maximum video bitrate of 1860 kbits/sec for NAL HRD parameters, and a CPB size of 4036458 bits (1550000 / 384000 * 1000 * 1000). The value of max-br MUST be greater than or equal to the value MaxBR for the signaled level given in Table A-1 of [H.264]. Wenger, et al Expires December 30, 2008 [Page 53] Internet-Draft RTP Payload Format for SVC Video June 2008 Senders MAY use this knowledge to send higher bitrate video as allowed in the level definition of SVC, to achieve improved video quality. Informative note: This parameter was added primarily to complement a similar codepoint in the ITU-T Recommendation H.245, so as to facilitate signaling gateway designs. No assumption can be made from the value of this parameter that the network is capable of handling such bit rates at any given time. In particular, no conclusion can be drawn that the signaled bit rate is possible under congestion control constraints. redundant-pic-cap: This parameter is as specified in RFC 3984. sprop-parameter-sets: This parameter MAY be used to convey any sequence parameter set, subset sequence parameter set and picture parameter set NAL units (herein referred to as the initial parameter set NAL units) that MUST be placed in the NAL unit stream to precede any other NAL units in decoding order by the receiver. The parameter MUST NOT be used to indicate codec capability in any capability exchange procedure. The value of the parameter is the base64 [RFC3548] representation of the initial parameter set NAL units as specified in sections 7.3.2.1, 7.3.2.2 and G.7.3.2.1 of [H.264]. The parameter sets are conveyed in decoding order, and no framing of the parameter set NAL units takes place. A comma is used to separate any pair of parameter sets in the list. Note that the number of bytes in a parameter set NAL unit is typically less than 10, but a picture parameter set NAL unit can contain several hundreds of bytes. Informative note: When several payload types are offered in the SDP Offer/Answer model, each with its own sprop- parameter-sets parameter, then the receiver cannot assume that those parameter sets do not use conflicting storage locations (i.e., identical values of parameter set identifiers). Therefore, a receiver should double-buffer all sprop-parameter-sets and make them available to the decoder instance that decodes a certain payload type. parameter-add: This parameter is as specified in RFC 3984. Wenger, et al Expires December 30, 2008 [Page 54] Internet-Draft RTP Payload Format for SVC Video June 2008 packetization-mode: This parameter is as specified in RFC 3984. sprop-interleaving-depth: This parameter is as specified in RFC 3984. sprop-deint-buf-req: This parameter is as specified in RFC 3984. deint-buf-cap: This parameter is as specified in RFC 3984. sprop-init-buf-time: This parameter is as specified in RFC 3984. sprop-max-don-diff: This parameter is as specified in RFC 3984. max-rcmd-nalu-size: This parameter is as specified in RFC 3984. pmode: This parameter signals the properties of a NAL unit stream carried in more than one RTP session using session multiplexing or the capabilities of a receiver implementation. When the value of pmode is equal to "NI-T", the NI-T mode MUST be used. When the value of pmode is equal to "NI-C", the NI-C mode MUST be used. When the value of pmode is equal to "NI- TC" or pmode is not present, the NI-TC mode MUST be used. When the value of pmode is equal to "I-C", the I-C mode MUST be used. The value of pmode MUST have one of the following tokens: "NI-T", "NI-C", "NI-TC", or "I-C". This parameter MUST NOT be present, when "packetization-mode" is present. sprop-mst-interleave-depth: This parameter MUST NOT be present when the value of pmode is equal to "NI-T". This parameter MUST be present when the value of pmode is equal to "NI-C", "NI-TC", or "I-C" or pmode is not present. This parameter signals the properties of a NAL unit stream carried in the current RTP session and the RTP sessions the current RTP session depends on. It is guaranteed that receivers can reconstruct NAL unit decoding order as specified in Subsection 8.1.2 of this memo when the remultiplexing buffer size is at least the value of sprop-mst-interleave- depth + 1 in terms of VCL NAL units. [Ed. (AE): But you don't Wenger, et al Expires December 30, 2008 [Page 55] Internet-Draft RTP Payload Format for SVC Video June 2008 say what's the property explicitly. Only what it guarantees. Is it interleaving depth?] The value of sprop-mst-interleave-depth MUST be an integer in the range of 0 to 32767, inclusive. sprop-remux-buf-req: This parameter MUST NOT be present when the value of pmode is equal to "NI-T". It MUST be present when pmode is not present or the value of pmode is equal to "NI-C", "NI-TC", or "I-C". sprop-remux-buf-req signals the required size of the remultiplexing buffer for the NAL unit stream carried in the current RTP session and the RTP sessions the current RTP session depends on. It is guaranteed that receivers can recover the decoding order of the received NAL units from the current RTP session and the RTP sessions the current RTP session depends on as specified in section 8.1.2, when the remultiplexing buffer size is at least the value of sprop- remux-buf-req in terms of bytes. The value of sprop-remux-buf-req MUST be an integer in the range of 0 to 4294967295, inclusive. remux-buf-cap: This parameter signals the capabilities of a receiver implementation and indicates the amount of remultiplexing buffer space in units of bytes that the receiver has available for recovering the NAL unit decoding order as specified in section 8.1.2. A receiver is able to handle any NAL unit stream for which the value of the sprop-remux-buf-req parameter is smaller than or equal to this parameter. If the parameter is not present, then a value of 0 MUST be used for remux-buf-cap. The value of remux-buf-cap MUST be an integer in the range of 0 to 4294967295, inclusive. sprop-remux-init-buf-time: This parameter MAY be used to signal the properties of a NAL unit stream carried in the current RTP session and the RTP sessions the current RTP session depends on. The parameter MUST NOT be present if pmode is not present or the value of pmode is equal to "NI-C", "NI-TC", or "I-C". The parameter signals the initial buffering time that a receiver MUST wait before starting to recover the NAL unit decoding order as specified in Subsection 8.1.2 of this memo. Wenger, et al Expires December 30, 2008 [Page 56] Internet-Draft RTP Payload Format for SVC Video June 2008 The parameter is coded as a non-negative base10 integer representation in clock ticks of a 90-kHz clock. If the parameter is not present, then no initial buffering time value is defined. Otherwise the value of sprop-remux-init-buf-time MUST be an integer in the range of 0 to 4294967295, inclusive. sprop-mst-max-don-diff: This parameter MAY be used to signal the properties of a NAL unit stream carried in the current RTP session and the RTP sessions the current RTP session depends on. It MUST NOT be used to signal transmitter or receiver or codec capabilities. The parameter MUST NOT be present if the value of pmode is equal to "NI-T". sprop-mst-max-don-diff is an integer in the range of 0 to 32767, inclusive. If sprop-mst-max-don-diff is not present, the value of the parameter is unspecified. sprop-mst-max-don-diff is calculated same as sprop-max-don- diff as specified in RFC 3984, with decoding order number being replaced by cross-session decoding order number. sprop-prebuf-size: This parameter MAY be present when the current RTP session depends on any other RTP session. This parameter MUST NOT be present when pmode is not present or the value of pmode is equal to "NI-C", "NI-TC", or "I-C". sprop-prebuf-size MAY signal the required size of the receiver buffer for the NAL unit stream per RTP session. This parameter may be useful to compensate the impact of inter-RTP session jitter, when the receiver buffer size is at least the value of sprop-prebuf- size in terms of bytes. Herein the NAL unit stream refers to the one consisting of all NAL units conveyed in the current RTP session. The value of sprop-prebuf-size MUST be an integer in the range of 0 to 4294967295, inclusive. Informative note: sprop-prebuf-size indicates the required size of the prebuffering receiver buffer only. When network jitter can occur, an appropriately sized jitter buffer has to be provisioned for as well. When a scalable bitstream is conveyed in more than one RTP session, and the sessions initiates at different time, the session initiation variation has also to be compensated by an appropriately sized buffer. [Ed. (AE): None of the sprop-prebuf-* parameters are mentioned anywhere else in the spec. Why are they needed?] Wenger, et al Expires December 30, 2008 [Page 57] Internet-Draft RTP Payload Format for SVC Video June 2008 sprop-prebuf-time: This parameter MAY be used to signal the properties of a NAL unit stream within a session multiplexing. Herein the NAL unit stream refers to the one consisting of all NAL units conveyed in the current RTP session. This parameter MUST NOT be present when pmode is not present or the value of pmode is equal to "NI-C", "NI-TC", or "I-C". The parameter signals the initial buffering time used for a receiver before starting to recover the NAL unit decoding order for more than one RTP session. The parameter is the maximum value of (transmission time of a NAL unit - decoding time of the NAL unit), assuming reliable and instantaneous transmission, the same timeline for transmission and decoding, and that decoding starts when the first packet arrives. The parameter is coded as a non-negative base10 integer representation in clock ticks of a 90-kHz clock. If the parameter is not present, then no initial buffering time value is defined. Otherwise the value of sprop-prebuf-time MUST be an integer in the range of 0 to 4294967295, inclusive. In addition to the signaled sprop-prebuf-time, receivers SHOULD take into account the transmission delay jitter buffering, including buffering for the delay jitter caused by mixers, translators, gateways, proxies, traffic-shapers, and other network elements. Yet another aspect receivers SHOULD take into account is the session initiation variation when a scalable bitstream is conveyed in more than one session, including buffering the variation. [Ed. (YkW): Need to discuss how inter-RTP session jitter should be handled in general, and how it works by using sprop- prebuf-size and sprop-prebuf-time.] sprop-scalability-info: This parameter MAY be used to convey the NAL unit containing the scalability information SEI message as specified in Annex G of [H.264]. This parameter MAY be used to signal the contained Layers of an SVC bitstream. The parameter MUST NOT be used to indicate codec capability in any capability exchange procedure. The value of the parameter is the base64 representation of the NAL unit containing the scalability information SEI message. If present, the NAL unit MUST contain only a scalability information SEI message. Wenger, et al Expires December 30, 2008 [Page 58] Internet-Draft RTP Payload Format for SVC Video June 2008 This parameter MAY be used in an offering or declarative SDP message to indicate what Layers can be provided. A receiver MAY indicate its choice of one Layer using the optional media type parameter scalable-layer-id. sprop-layer-range: This parameter MAY be used to signal two sets of the layer identification values of the lowest and highest operation points conveyed in the RTP session. Each set is a base16 representation of a three-character value, with the first character representing DID, the second character representing QID, and the third character representing TID. The two sets are comma separated. Let DIDl and DIDh be the lowest DID value and the highest DID value, respectively, among all the NAL units conveyed in the RTP session. Let QIDl and TIDl be the lowest QID value and the lowest TID value, respectively, among all the NAL units that are conveyed in the RTP session and that have DID equal to DIDl. Let QIDh and TIDh be the highest QID value and the highest TID value, respectively, among all the NAL units that are conveyed in the RTP session and that have DID equal to DIDh. The first set indicates the DID, QID and TID values of the lowest operation point, for which the DID, QID and TID values are equal to DIDl, QIDl, and TIDl, respectively. The second set indicates the DID, QID and TID values of the highest operation point, for which the DID, QID and TID values are equal to DIDh, QIDh, and TIDh, respectively. scalable-layer-id: This parameter MAY be used to signal a receiver's choice of the offers or declared operation points or layers using sprop- scalability-info. The value of scalable-layer-id is a base16 representation of the layer_id[ i ] syntax element in the scalability information SEI message as specified in [H.264]. [Ed. (TS): That is, a SDP capable receiver/middle-box must decode the sprop-scalabiltiy-info syntax, which is not specified in this memo, to select a scalable-layer-id. This is currently not addressed in the offer answer section!] sprop-spatial-resolution: [Ed. (??): I know that framerate and bitrate SDP parameters are already available, but failed to find a spatial resolution SDP parameter. It would be good if this is already defined. Otherwise, it would be better to be defined somewhere else because it is a generic parameter.] Wenger, et al Expires December 30, 2008 [Page 59] Internet-Draft RTP Payload Format for SVC Video June 2008 This parameter MAY be used to indicate the property of a stream or the capability of a receiver or sender implementation. The value is a base16 of the width and height of the spatial resolution, in pixels, separated by a comma. [Edt. Note (TS): Shouldn't this be a generic SDP parameter?] Encoding considerations: This type is only defined for transfer via RTP (RFC 3550). Security considerations: See Section 10 of RFC XXXX. Public specification: Please refer to Section 14 of RFC XXXX. Additional information: None File extensions: none Macintosh file type code: none Object identifier or OID: none Person & email address to contact for further information: Intended usage: COMMON Author: Change controller: IETF Audio/Video Transport working group delegated from the IESG. 9.2 SDP Parameters [Ed. (??): For agreeing on a Layer or OP in unicast, an SDP can contain multiple m lines with bitrate, framerate and spatial resolution parameters available, in addition to sprop-scalability- info. The receive can select one of the m lines, or, for operation points that are not included in the m lines, one of the "scalable layers" specified by sprop-scalabiltiy-info using scalable-layer-id. For layered multicast, then the grouping signaling in I-D.ietf- mmusic-decoding-dependency is needed. Wenger, et al Expires December 30, 2008 [Page 60] Internet-Draft RTP Payload Format for SVC Video June 2008 The above would conveniently support also the normal ROI use cases (with a few ROIs each indicated as a "scalable layer") but not the interactive ROI use cases. The quality layer using priority_id use cases are not supported either. That would need one more optional media type parameter, to identify a quality layer. The lightweight transcoding use cases should be supported well by using (multiple) normal AVC SDP offering messages.] 9.2.1 Mapping of Payload Type Parameters to SDP The media type video/H264-SVC string is mapped to fields in the Session Description Protocol (SDP) as follows: o The media name in the "m=" line of SDP MUST be video. o The encoding name in the "a=rtpmap" line of SDP MUST be H264-SVC (the media subtype). o The clock rate in the "a=rtpmap" line MUST be 90000. o The OPTIONAL parameters "profile-level-id", "max-mbps", "max-fs", "max-cpb", "max-dpb", "max-br", "redundant-pic-cap", "sprop- parameter-sets", "parameter-add", "packetization-mode", "sprop- interleaving-depth", "deint-buf-cap", "sprop-deint-buf-req", "sprop-init-buf-time", "sprop-max-don-diff", "max-rcmd-nalu- size", "pmode", "sprop-mst-interleave-depth", "sprop-remux-buf- req", "remux-buf-cap", "sprop-remux-init-buf-time", "sprop-mst- max-don-diff", "sprop-prebuf-size", "sprop-prebuf-time", "sprop- layer-range", "sprop-scalability-info", "scalable-layer-id", and "sprop-spatial-resolution", when present, MUST be included in the "a=fmtp" line of SDP. These parameters are expressed as a media type string, in the form of a semicolon separated list of parameter=value pairs. 9.2.2 Usage with the SDP Offer/Answer Model When H.264 or SVC is offered over RTP using SDP in an Offer/Answer model [RFC3264] for negotiation for unicast usage, the following limitations and rules apply: o The parameters identifying a media format configuration for H.264 or SVC are "profile-level-id", "packetization-mode", and, if required by "packetization-mode", "sprop-deint-buf-req". These three parameters MUST be used symmetrically; i.e., the answerer MUST either maintain all configuration parameters or remove the media format (payload type) completely, if one or more of the parameter values are not supported. Wenger, et al Expires December 30, 2008 [Page 61] Internet-Draft RTP Payload Format for SVC Video June 2008 Informative note: The requirement for symmetric use applies only for the above three parameters and not for the other stream properties and capability parameters. To simplify handling and matching of these configurations, the same RTP payload type number used in the offer SHOULD also be used in the answer, as specified in [RFC3264]. An answer MUST NOT contain a payload type number used in the offer unless the configuration ("profile-level-id", "packetization-mode", and, if present, "sprop-deint-buf-req") is the same as in the offer. Informative note: An offerer, when receiving the answer, has to compare payload types not declared in the offer based on media type (i.e., video/H264-SVC) and the above three parameters with any payload types it has already declared, in order to determine whether the configuration in question is new or equivalent to a configuration already offered. An answerer MAY select from the layers offered in the "sprop- scalability-information" parameter by including "scalable-layer- id" or "sprop-layer-range" in the answer.[Edt. Note: do we need to additionally define behavior with snd/rcvonly parameter?] o The parameters "sprop-parameter-sets", "sprop-deint-buf-req", "sprop-interleaving-depth", "sprop-max-don-diff", "sprop-init- buf-time", "sprop-prebuf-size", "sprop-prebuf-time", "sprop- scalability-information", "sprop-layer-range" describe the properties of the NAL unit stream that the offerer or answerer is sending for this media format configuration. This differs from the normal usage of the Offer/Answer parameters: normally such parameters declare the properties of the stream that the offerer or the answerer is able to receive. When dealing with H.264 or SVC, the offerer assumes that the answerer will be able to receive media encoded using the configuration being offered. Informative note: The above parameters apply for any stream sent by the declaring entity with the same configuration; i.e., they are dependent on their source. Rather then being bound to the payload type, the values may have to be applied to another payload type when being sent, as they apply for the configuration. Wenger, et al Expires December 30, 2008 [Page 62] Internet-Draft RTP Payload Format for SVC Video June 2008 o The capability parameters ("max-mbps", "max-fs", "max-cpb", "max-dpb", "max-br", ,"redundant-pic-cap", "max-rcmd-nalu-size") MAY be used to declare further capabilities. Their interpretation depends on the direction attribute. When the direction attribute is sendonly, then the parameters describe the limits of the RTP packets and the NAL unit stream that the sender is capable of producing. When the direction attribute is sendrecv or recvonly, then the parameters describe the limitations of what the receiver accepts. o As specified above, an offerer has to include the size of the deinterleaving buffer in the offer for an interleaved H.264 or SVC stream. To enable the offerer and answerer to inform each other about their capabilities for deinterleaving buffering, both parties are RECOMMENDED to include "deint-buf-cap". This information MAY be used when the value for "sprop-deint-buf-req" is selected in a second round of offer and answer. For interleaved streams, it is also RECOMMENDED to consider offering multiple payload types with different buffering requirements when the capabilities of the receiver are unknown. o The "sprop-parameter-sets" parameter is used as described above. In addition, an answerer MUST maintain all parameter sets received in the offer in its answer. Depending on the value of the "parameter-add" parameter, different rules apply: If "parameter-add" is false (0), the answer MUST NOT add any additional parameter sets. If "parameter-add" is true (1), the answerer, in its answer, MAY add additional parameter sets to the "sprop-parameter-sets" parameter. The answerer MUST also, independent of the value of "parameter-add", accept to receive a video stream using the sprop-parameter-sets it declared in the answer. Informative note: care must be taken when parameter sets are added not to cause overwriting of already transmitted parameter sets by using conflicting parameter set identifiers. For streams being delivered over multicast, the following rules apply in addition: o The stream properties parameters ("sprop-parameter-sets", "sprop- deint-buf-req", "sprop-interleaving-depth", "sprop-max-don-diff", "sprop-init-buf-time", "sprop-prebuf-size", "sprop-prebuf-time", "sprop-scalability-information", and "sprop-layer-range") MUST NOT be changed by the answerer. Thus, a payload type can either be accepted unaltered or removed. Wenger, et al Expires December 30, 2008 [Page 63] Internet-Draft RTP Payload Format for SVC Video June 2008 o The receiver capability parameters "max-mbps", "max-fs", "max- cpb", "max-dpb", "max-br", and "max-rcmd-nalu-size" MUST be supported by the answerer for all streams declared as sendrecv or recvonly; otherwise, one of the following actions MUST be performed: the media format is removed, or the session rejected. o The receiver capability parameter redundant-pic-cap SHOULD be supported by the answerer for all streams declared as sendrecv or recvonly as follows: The answerer SHOULD NOT include redundant coded pictures in the transmitted stream if the offerer indicated redundant-pic-cap equal to 0. Otherwise (when redundant_pic_cap is equal to 1), it is beyond the scope of this memo to recommend how the answerer should use redundant coded pictures. Below are the complete lists of how the different parameters shall be interpreted in the different combinations of offer or answer and direction attribute. o In offers and answers for which "a=sendrecv" or no direction attribute is used, or in offers and answers for which "a=recvonly" is used, the following interpretation of the parameters MUST be used. Declaring actual configuration or properties for receiving: - profile-level-id - packetization-mode Declaring actual properties of the stream to be sent (applicable only when "a=sendrecv" or no direction attribute is used): - sprop-deint-buf-req - sprop-interleaving-depth - sprop-parameter-sets - sprop-max-don-diff - sprop-init-buf-time - sprop-prebuf-size - sprop-prebuf-time - sprop-scalability-information - sprop-layer-range - scalable-layer-id Declaring receiver implementation capabilities: - max-mbps - max-fs - max-cpb Wenger, et al Expires December 30, 2008 [Page 64] Internet-Draft RTP Payload Format for SVC Video June 2008 - max-dpb - max-br - redundant-pic-cap - deint-buf-cap - max-rcmd-nalu-size Declaring how Offer/Answer negotiation shall be performed: - parameter-add o In an offer or answer for which the direction attribute "a=sendonly" is included for the media stream, the following interpretation of the parameters MUST be used: Declaring actual configuration and properties of stream proposed to be sent: - profile-level-id - packetization-mode - sprop-deint-buf-req - sprop-max-don-diff - sprop-init-buf-time - sprop-parameter-sets - sprop-interleaving-depth - sprop-prebuf-size - sprop-prebuf-time - sprop-scalability-information - sprop-layer-range - sprop-spatial-resoltuion Declaring how Offer/Answer negotiation shall be performed: - parameter-add Furthermore, the following considerations are necessary: o Parameters used for declaring receiver capabilities are in general downgradable; i.e., they express the upper limit for a sender's possible behavior. Thus a sender MAY select to set its encoder using only lower/lesser or equal values of these parameters. "sprop-parameter-sets" MUST NOT be used in a sender's declaration of its capabilities, as the limits of the values that are carried inside the parameter sets are implicit with the profile and level used. Wenger, et al Expires December 30, 2008 [Page 65] Internet-Draft RTP Payload Format for SVC Video June 2008 o Parameters declaring a configuration point are not downgradable, with the exception of the level part of the "profile-level-id" parameter. This expresses values a receiver expects to be used and must be used verbatim on the sender side. o When a sender's capabilities are declared, and non-downgradable parameters are used in this declaration, then these parameters express a configuration that is acceptable. In order to achieve high interoperability levels, it is often advisable to offer multiple alternative configurations; e.g., for the packetization mode. It is impossible to offer multiple configurations in a single payload type. Thus, when multiple configuration offers are made, each offer requires its own RTP payload type associated with the offer. o A receiver SHOULD understand all MIME parameters, even if it only supports a subset of the payload format's functionality. This ensures that a receiver is capable of understanding when an offer to receive media can be downgraded to what is supported by receiver of the offer. o An answerer MAY extend the offer with additional media format configurations. However, to enable their usage, in most cases a second offer is required from the offerer to provide the stream properties parameters that the media sender will use. This also has the effect that the offerer has to be able to receive this media format configuration, not only to send it. o If an offerer wishes to have non-symmetric capabilities between sending and receiving, the offerer has to offer different RTP sessions; i.e., different media lines declared as "recvonly" and "sendonly", respectively. This may have further implications on the system. 9.2.3 Usage with Multi-Source Transmission If MST is used, the rules on signaling media decoding dependency in SDP as defined in [I-D.ietf-mmusic-decoding-dependency] apply. 9.2.4 Usage in Declarative Session Descriptions When H.264 or SVC over RTP is offered with SDP in a declarative style, as in RTSP [RFC2326] or SAP [RFC2974], the following considerations are necessary. Wenger, et al Expires December 30, 2008 [Page 66] Internet-Draft RTP Payload Format for SVC Video June 2008 o All parameters capable of indicating the properties of both a NAL unit stream and a receiver are used to indicate the properties of a NAL unit stream. For example, in this case, the parameter "profile-level-id" declares the values used by the stream, instead of the capabilities of the sender. This results in that the following interpretation of the parameters MUST be used: Declaring actual configuration or properties: - profile-level-id - sprop-parameter-sets - packetization-mode - sprop-interleaving-depth - sprop-deint-buf-req - sprop-max-don-diff - sprop-init-buf-time - sprop-prebuf-size - sprop-prebuf-time - sprop-layer-range - sprop-spatial-resolution - sprop-scalability-info Not usable: - max-mbps - max-fs - max-cpb - max-dpb - max-br - redundant-pic-cap - max-rcmd-nalu-size - parameter-add - deint-buf-cap - scalable_layer_id o A receiver of the SDP is required to support all parameters and values of the parameters provided; otherwise, the receiver MUST reject (RTSP) or not participate in (SAP) the session. It falls on the creator of the session to use values that are expected to be supported by the receiving application. 9.3 Examples 9.3.1 Example for Offering A Single SVC Session Offerer -> Answerer SDP message: Wenger, et al Expires December 30, 2008 [Page 67] Internet-Draft RTP Payload Format for SVC Video June 2008 v=0 o=jdoe 2890844526 2890842807 IN IP4 192.0.2.12 s=SVC SDP example i=SVC Scalable Video Coding session t=2873397496 2873404696 m=video 20000 RTP/AVP 96 97 98 a=rtpmap:96 H264/90000 a=fmtp:96 profile-level-id=4d400a; packetization-mode=1; sprop-parameter-sets=Z01ACprLFicg,aP4Eag==; a=rtpmap:97 H264-SVC/90000 a=fmtp:97 profile-level-id=53000c; packetization-mode=1; sprop-parameter-sets=Z01ACprLFicg,Z1MADEsA1NZYWCWQ,aP4Eag==,aEvgR qA=,aGvgRiA=; a=rtpmap:98 H264-SVC/90000 a=fmtp:98 profile-level-id=53000c; packetization-mode=2; init-buf-time=156320; sprop-parameter-sets=Z01ACprLFicg,Z1MADEsA1NZYWCWQ,aP4Eag==,aEvgR qA=,aGvgRiA=; 9.3.2 Example for Offering Session Multiplexing Offerer -> Answerer SDP message: v=0 o=jdoe 2890844526 2890842807 IN IP4 192.0.2.12 s=SVC Scalable Video Coding session i=SDP is an Offer for a session offered by a transcoding entity t=2873397496 2873404696 a=group:DDP 1 2 3 m=video 20000 RTP/AVP 96 97 98 a=rtpmap:96 H264/90000 a=fmtp:96 profile-level-id=4d400a; packetization-mode=0; sprop-parameter-sets=Z01ACprLFicg,aP4Eag==; a=rtpmap:97 H264/90000 a=fmtp:97 profile-level-id=53000c; packetization-mode=1; sprop-parameter-sets=Z01ACprLFicg,Z1MADEsA1NZYWCWQ,aP4Eag==,aEvgR qA=,aGvgRiA=; a=rtpmap:98 H264/90000 a=fmtp:98 profile-level-id=53000c; packetization-mode=2; init-buf-time=156320; sprop-parameter-sets=Z01ACprLFicg,Z1MADEsA1NZYWCWQ,aP4Eag==,aEvgR qA=,aGvgRiA=; a=mid:1 m=video 20002 RTP/AVP 99 100 a=rtpmap:99 H264-SVC/90000 a=fmtp:99 profile-level-id=53000c; pmode=NI-T; sprop-parameter-sets=Z01ACprLFicg,Z1MADEsA1NZYWCWQ,aP4Eag==,aEvgR Wenger, et al Expires December 30, 2008 [Page 68] Internet-Draft RTP Payload Format for SVC Video June 2008 qA=,aGvgRiA=; a=rtpmap:100 H264-SVC/90000 a=fmtp:100 profile-level-id=53000c; pmode=I-C; sprop-parameter-sets=Z01ACprLFicg,Z1MADEsA1NZYWCWQ,aP4Eag==,aEvgR qA=,aGvgRiA=; a=mid:2 a=depend:99 lay 1:96,97; 100 lay 1:98 m=video 20004 RTP/AVP 101 a=rtpmap:101 H264-SVC/90000 a=fmtp:101 profile-level-id=53000c; pmode=NI-TC; sprop-parameter-sets=Z01ACprLFicg,Z1MADEsA1NZYWCWQ,aP4Eag==,aEvgR qA=,aGvgRiA=; a=mid:3 a=depend:101 lay 1:96,97 2:99 9.4 Parameter Set Considerations Please see Section 8.4 of [RFC3984]. 10. Security Considerations Section 9 of [RFC3984] applies. Additionally, the following applies. Decoders MUST exercise caution with respect to the handling of reserved NAL unit types and reserved SEI messages, particularly if they contain active elements, and MUST restrict their domain of applicability to the presentation containing the stream. The safest way is to simply discard these NAL units and SEI messages. When integrity protection is applied, care MUST be taken that the stream being transported may be scalable; hence a receiver may be able to access only part of the entire stream. Informative note: Other security aspects, including confidentiality, authentication, and denial-of-service threat, for SVC are similar as H.264/AVC, as discussed in Section 9 of [RFC3984]. 11. Congestion Control Within any given RTP session carrying payload according to this specification, the provisions of section 12 of [RFC3984] apply. Reducing the session bandwidth is possible by one or more of the following means, listed in an order that, in most cases, will assure the least negative impact to the user experience: Wenger, et al Expires December 30, 2008 [Page 69] Internet-Draft RTP Payload Format for SVC Video June 2008 a) within the highest Layer identified by the DIDfield, utilize the TID and/or QID fields in the NAL unit header to drop NAL units with lower importance for the decoding process or human perception. b) drop all NAL units belonging to the highest enhancement Layer as identified by the highest DID value. c) dropping NAL units according to their importance for the decoding process, as indicated by the fields in the NAL unit header of the NAL units or in the prefix NAL units. d) dropping NAL units or entire packets not according to the aforementioned rules (media-unaware stream thinning). This results in the reception of a non-compliant bitstream and, most likely, in very annoying artifacts Informative note: The discussion above is centered on NAL units and not on packets, primarily because that is the level where senders can meaningfully manipulate the scalable bitstream. The mapping of NAL units to RTP packets is fairly flexible when using aggregation packets. Depending on the nature of the congestion control algorithm, the "dimension" of congestion measurement (packet count or bitrate) and reaction to it (reducing packet count or bitrate or both) can be adjusted accordingly. All aforementioned means are available to the RTP sender, regardless whether that sender is located in the sending endpoint or in a mixer based MANE. When a translator-based MANE is employed, then the MANE MAY manipulate the session only on the MANE's outgoing path, so that the sensed end-to-end congestion falls within the permissible envelope. As all translators, in this case the MANE needs to rewrite RTCP RRs to reflect the manipulations it has performed on the session. Informative note: Applications MAY also implement, in addition or separately, other congestion control mechanisms, e.g. as described in [RFC3450] and [Yan]. 12. IANA Consideration [Edt. Note: A new media type should be registered from IANA.] Wenger, et al Expires December 30, 2008 [Page 70] Internet-Draft RTP Payload Format for SVC Video June 2008 13. Informative Appendix: Application Examples [Ed. (AE): I think this whole section does not add any real value, is outdated, and should be eliminated. In particular, the application scenario of the only currently shipping SVC product is not even listed here.] 13.1 Introduction Scalable video coding is a concept that has been around since at least MPEG-2 [MPEG2], which goes back as early as 1993. Nevertheless, it has never gained wide acceptance; perhaps partly because applications didn't materialize in the form envisioned during standardization. ISO/IEC MPEG and ITU-T VCEG, respectively, performed a requirement analysis for the SVC project. Dozens of scenarios have been studied. While some of the scenarios appear not to follow the most basic design principles of the Internet, e.g. as discussed in section 13.5 -- and are therefore not appropriate for IETF standardization -- others are clearly in the scope of IETF work. Of these, this draft chooses the following subset for immediate consideration. The MPEG and VCEG requirement documents are available in [JVT-N026] and [JVT-N027], respectively. With these remarks, we now introduce three main application scenarios that we consider relevant, and that are implementable with this specification. 13.2 Layered Multicast This well-understood form of the use of layered coding [McCanne] implies that all layers are individually conveyed in their own RTP packet streams, each carried in its own RTP session using the IP (multicast) address and port number as the single demultiplexing point. Receivers "tune" into the layers by subscribing to the IP multicast, normally by using IGMP [IGMP]. Depending on the application scenario, it is also possible to convey a number of layers in one RTP session, when finer operation points within the subset of layers are not needed. Layered multicast has the great advantage of simplicity and easy implementation. However, it has also the great disadvantage of utilizing many different transport addresses. While we consider this not to be a major problem for a professionally maintained content server, receiving client endpoints need to open many ports to IP multicast addresses in their firewalls. This is a practical Wenger, et al Expires December 30, 2008 [Page 71] Internet-Draft RTP Payload Format for SVC Video June 2008 problem from a firewall and network address translation (NAT) viewpoint. Furthermore, even today IP multicast is not as widely deployed as many wish. We consider layered multicast an important application scenario for the following reasons. First, it is well understood and the implementation constraints are well known. Second, there may well be large scale IP networks outside the immediate Internet context that may wish to employ layered multicast in the future. One possible example could be a combination of content creation and core-network distribution for the various mobile TV services, e.g. those being developed by 3GPP (MBMS) [MBMS] and DVB (DVB-H) [DVB-H]. 13.3 Streaming of an SVC Scalable Stream In this scenario, a streaming server has a repository of stored SVC coded layers for a given content. At the time of streaming, and according to the capabilities, connectivity, and congestion situation of the client(s), the streaming server generates and serves a scalable stream. Both unicast and multicast serving is possible. At the same time, the streaming server may use the same repository of stored layers to compose different streams (with a different set of layers) intended for other audiences. As every endpoint receives only a single SVC RTP session, the number of firewall pinholes can be optimized to one. The main difference between this scenario and straightforward simulcasting lies in the architecture and the requirements of the streaming server, and is therefore out of the scope of IETF standardization. However, compelling arguments can be made why such a streaming server design makes sense. One possible argument is related to storage space and channel bandwidth. Another is bandwidth adaptability without transcoding -- a considerable advantage in a congestion controlled network. When the streaming server learns about congestion, it can reduce sending bitrate by choosing fewer layers, when composing the layered stream; see section 11. SVC is designed to gracefully support both bandwidth rampdown and bandwidth rampup with a considerable dynamic range. This payload format is designed to allow for bandwidth flexibility in the mentioned sense. While, in theory, a transcoding step could achieve a similar dynamic range, the computational demands are impractically high and video quality is typically lowered -- therefore, few (if any) streaming servers implement full transcoding. Wenger, et al Expires December 30, 2008 [Page 72] Internet-Draft RTP Payload Format for SVC Video June 2008 13.4 Multicast to MANE, SVC Scalable Stream to Endpoint This scenario is a bit more complex, and designed to optimize the network traffic in a core network, while still requiring only a single pinhole in the endpoint's firewall. One of its key applications is the mobile TV market. Consider a large private IP network, e.g. the core network of 3GPP. Streaming servers within this core network can be assumed to be professionally maintained. We assume that these servers can have many ports open to the network and that layered multicast is a real option. Therefore, we assume that the streaming server multicasts SVC scalable layers, instead of simulcasting different representations of the same content at different bit rates. Also consider many endpoints of different classes. Some of these endpoints may not have the processing power or the display size to meaningfully decode all layers; others may have these capabilities. Users of some endpoints may not wish to pay for high quality and are happy with a base service, which may be cheaper or even free. Other users are willing to pay for high quality. Finally, some connected users may have a bandwidth problem in that they can't receive the bandwidth they would want to receive -- be it through congestion, connectivity, change of service quality, or for whatever other reasons. However, all these users have in common that they don't want to be exposed too much, and therefore the number of firewall pinholes need to be small. This situation can be handled best by introducing middleboxes close to the edge of the core network, which receive the layered multicast streams and compose the single SVC scalable bit stream according to the needs of the endpoint connected. These middleboxes are called MANEs throughout this specification. In practice, we envision the MANE to be part of (or at least physically and topologically close to) the base station of a mobile network, where all the signaling and media traffic necessarily are multiplexed on the same physical link. This is why we do not worry too much about decomposition aspects of the MANE as such. MANEs necessarily need to be fairly complex devices. They certainly need to understand the signaling, so, for example, to associate the PT octet in the RTP header with the SVC payload type. A MANE may aggregate multiple RTP streams, possibly from multiple RTP sessions, thus to reduce the number of firewall pinholes required at the endpoints. This type of MANEs is conceptually easy to implement and can offer powerful features, primarily because it Wenger, et al Expires December 30, 2008 [Page 73] Internet-Draft RTP Payload Format for SVC Video June 2008 necessarily can "see" the payload (including the RTP payload headers), utilize the wealth of layering information available therein, and manipulate it. While such an MANE operation in its most trivial form (combining multiple RTP packet streams into a single one) can be implemented comparatively simply -- reordering the incoming packets according to the DON and sending them in the appropriate order -- more complex forms can also be envisioned. For example, a MANE can be optimizing the outgoing RTP stream to the MTU size of the outgoing path by utilizing the aggregation and fragmentation mechanisms of this memo. A MANE can also perform stream thinning, so to adhere to congestion control principles as discussed in section 11. While the implementation of the forward (media) channel of such a MANE appears to be comparatively simple, the need to rewrite RTCP RRs makes even such a MANE a complex device. While the implementation complexity of either case of a MANE, as discussed above, is fairly high, the computational demands are comparatively low. In particular, SVC and/or this specification contain means to easily generate the correct inter-layer decoding order of NAL units. No serious bit-oriented processing is required and no significant state information (beyond that of the signaling and perhaps the SVC sequence parameter sets) need to be kept. 13.5 Scenarios Currently Not Considered Remarks have been made that the current draft does not take into consideration at least one application scenario which some JVT folks considered important. In particular, their idea was to make the RTP payload format (or the media stream itself) self-contained enough that a stateless, non-signaling-aware device can "thin" an RTP session to meet the bandwidth demands of the endpoint. They called this device a "Router" or "Gateway", and sometimes a MANE. Obviously, it's not a Router or Gateway in the IETF sense. To distinguish it from a MANE as defined in RFC 3984 and in this specification, let's call it an MDfH (Magic Device from Heaven). To simplify discussions, let's assume point-to-point traffic only. The endpoint has a signaling relationship with the streaming server, but it is known that the MDfH is somewhere in the media path (e.g. because the physical network topology ensures this). It has been requested, at least implicitly through MPEG's and JVT's requirements document, that the MDfH should be capable to intercept the SVC scalable bit stream, modify it by dropping packets or parts thereof, and forwarding the resulting packet stream to the receiving Wenger, et al Expires December 30, 2008 [Page 74] Internet-Draft RTP Payload Format for SVC Video June 2008 endpoint. It has been requested that this payload specification contains protocol elements facilitating such an operation, and the argument has been made that the NRI field of RFC 3984 serves exactly the same purpose. The authors of this I-D do not consider the scenario above to be aligned with the most basic design philosophies the IETF follows, and therefore have not addressed the comments made (except through this section). In particular, we see the following problems with the MDfH approach): o As the very minimum, the MDfH would need to know which RTP streams are carrying SVC. We don't see how this could be accomplished but by using a static payload type. None of the IETF defined RTP profiles envision static payload types for SVC, and even the de- facto profiles developed by some application standard organizations (3GPP for example) do not use this outdated concept. Therefore, the MDfH necessarily needs to be at least "listening" to the signaling. o If the RTP packet payload were encrypted, it would be impossible to interpret the payload header and/or the first bytes of the media stream. We understand that there are crypto schemes under discussion that encrypt only the last n bytes of an RTP payload, but we are more than unsure that this is fully in line with the IETF's security vision. Even if the above two problems would have been overcome through standardization outside of the IETF, we still foresee serious design flaws: o An MDfH can't simply dump RTP packets it doesn't want to forward. It either needs to act as a full RTP Translator (implying that it rewrites RTCP RRs and such), or it needs to patch the RTP sequence numbers to fulfill the RTP specification. Not doing either would, for the receiver, look like the gaps in the sequence numbers occurred due to unintentional erasures, which has interesting effects on congestion control (if implemented), will break pretty much every meta-payload ever developed, and so on. (Many more points could be made here). In summary, based on our current knowledge we are not willing to specify protocol mechanisms that support an operation point that has so little in common with classic RTP use. Wenger, et al Expires December 30, 2008 [Page 75] Internet-Draft RTP Payload Format for SVC Video June 2008 13.6 SSRC Multiplexing The authors have played with the idea of introducing SSRC multiplexing, i.e. allowing sending multiple RTP packet streams containing layers in the same RTP session, differentiated by SSRC values. Our intention was to minimize the number of firewall pinholes in an endpoint to one, by using MANEs to aggregate multiple outgoing sessions stemming from a server into a single session (with SSRC multiplexed packet streams). We were hoping that would be feasible even with encrypted packets in an SRTP context. While an implementation along these lines indeed appears to be feasible for the forward media path, the RTCP RR rewrite cannot be implemented in the way necessary for this scheme to work. This relates to the need to authenticate the RTCP RRs as per SRTP [RFC3711]. While the RTCP RR itself does not need to be rewritten by the scheme we envisioned, its transport addresses needs to be manipulated. This, in turn, is incompatible with the mandatory authentication of RTCP RRs. As a result, there would be a requirement that a MANE needs to be in the RTCP security context of the sessions, which was not envisioned in our use case. As the envisioned use case cannot be implemented, we refrained to add the considerable document complexity to support SSRC multiplexing herein. 14. References 14.1 Normative References [H.264] ITU-T Recommendation H.264, "Advanced video coding for rd generic audiovisual services", 3 Edition, November 2007. [I-D.ietf-mmusic-decoding-dependency] Schierl, T., and Wenger, S., "Signaling media decoding dependency in Session Description Protocol (SDP)", draft-ietf-mmusic-decoding- dependency-02 (work in progress), May 2008. [MPEG4-10] ISO/IEC International Standard 14496-10:2005. [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. [RFC3264] Rosenberg, J. and H. Schulzrinne, "An Offer/Answer Model With Session Description Protocol (SDP)", RFC 3264, June 2002. Wenger, et al Expires December 30, 2008 [Page 76] Internet-Draft RTP Payload Format for SVC Video June 2008 [RFC3548] Josefsson, S., "The Base16, Base32, and Base64 Data Encodings", RFC 3548, July 2003. [RFC3550] Schulzrinne, H., Casner, S., Frederick, R., and Jacobson, V., "RTP: A Transport Protocol for Real-Time Applications", STD 64, RFC 3550, July 2003. [RFC3984] Wenger, S., Hannuksela, M., Stockhammer, T., Westerlund,M., and Singer, D., "RTP Payload Format for H.264 Video", RFC 3984, February 2005. [RFC4566] Handley, M., Jacobson, V., and Perkins, C., "SDP: Session Description Protocol", RFC 4566, July 2006. 14.2 Informative References [DVB-H] DVB - Digital Video Broadcasting (DVB); DVB-H Implementation Guidelines, ETSI TR 102 377, 2005. [H.241] ITU-T Rec. H.241, "Extended video procedures and control signals for H.300-series terminals", May 2006. [IGMP] Cain, B., Deering S., Kovenlas, I., Fenner, B., and Thyagarajan, A., "Internet Group Management Protocol, Version 3", RFC 3376, October 2002. [JVT-N026] Ohm J.-R., Koenen, R., and Chiariglione, L. (ed.), "SVC requirements specified by MPEG (ISO/IEC JTC1 SC29 WG11)", JVT-N026, available from http://ftp3.itu.ch/av-arch/jvt- site/2005_01_HongKongGeneva/JVT-N026.doc, Hong Kong, China, January 2005. [JVT-N027] Sullivan, G., and Wiegand, T. (ed.), "SVC requirements specified by VCEG (ITU-T SG16 Q.6)", JVT-N027, available from http://ftp3.itu.ch/av-arch/jvt- site/2005_01_HongKongGeneva/JVT-N027.doc, Hong Kong, China, January 2005. [McCanne] McCanne, S., Jacobson, V., and Vetterli, M., "Receiver- driven layered multicast", in Proc. of ACM SIGCOMM'96, pages 117--130, Stanford, CA, August 1996. [MBMS] 3GPP - Technical Specification Group Services and System Aspects; Multimedia Broadcast/Multicast Service (MBMS); Protocols and codecs (Release 6), December 2005. [MPEG2] ISO/IEC International Standard 13818-2:1993. Wenger, et al Expires December 30, 2008 [Page 77] Internet-Draft RTP Payload Format for SVC Video June 2008 [RFC2326] Schulzrinne, H., Rao, A., and R. Lanphier, "Real Time Streaming Protocol (RTSP)", RFC 2326, April 1998. [RFC2974] Handley, M., Perkins, C., and E. Whelan, "Session Announcement Protocol", RFC 2974, October 2000. [RFC3450] Luby, M., Gemmell, J., Vicisano, L., Rizzo, L., and Crowcroft, J., "Asynchronous layered coding (ALC) protocol instantiation", RFC 3450, December 2002. [RFC3711] Baugher, M., McGrew, D, Naslund, M., Carrara, E., and Norrman, K., "The secure real-time transport protocol (SRTP)", RFC 3711, March 2004. [Yan] Yan, J., Katrinis, K., May, M., and Plattner, R., "Media- And TCP-friendly congestion control for scalable video streams", in IEEE Trans. Multimedia, pages 196--206, April 2006. 15. Authors' Addresses Stephan Wenger Nokia 955 Page Mill Road Palo Alto, CA 94304 USA Phone: +1-650-862-7368 EMail: stewe@stewe.org Ye-Kui Wang Nokia Research Center P.O. Box 100 33721 Tampere Finland Phone: +358-50-466-7004 EMail: ye-kui.wang@nokia.com Thomas Schierl Fraunhofer HHI Einsteinufer 37 D-10587 Berlin Wenger, et al Expires December 30, 2008 [Page 78] Internet-Draft RTP Payload Format for SVC Video June 2008 Germany Phone: +49-30-31002-227 Email: schierl@hhi.fhg.de Alex Eleftheriadis Vidyo, Inc. 433 Hackensack Ave. Hackensack, NJ 07601 USA Phone: +1-201-467-5135 Email: alex@vidyo.com Intellectual Property Statement The IETF takes no position regarding the validity or scope of any Intellectual Property Rights or other rights that might be claimed to pertain to the implementation or use of the technology described in this document or the extent to which any license under such rights might or might not be available; nor does it represent that it has made any independent effort to identify any such rights. Information on the procedures with respect to rights in RFC documents can be found in BCP 78 and BCP 79. Copies of IPR disclosures made to the IETF Secretariat and any assurances of licenses to be made available, or the result of an attempt made to obtain a general license or permission for the use of such proprietary rights by implementers or users of this specification can be obtained from the IETF on-line IPR repository at http://www.ietf.org/ipr. The IETF invites any interested party to bring to its attention any copyrights, patents or patent applications, or other proprietary rights that may cover technology that may be required to implement this standard. Please address the information to the IETF at ietf-ipr@ietf.org. Disclaimer of Validity This document and the information contained herein are provided on an "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY, THE IETF TRUST AND THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL Wenger, et al Expires December 30, 2008 [Page 79] Internet-Draft RTP Payload Format for SVC Video June 2008 WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Copyright Statement Copyright (C) The IETF Trust (2008). This document is subject to the rights, licenses and restrictions contained in BCP 78, and except as set forth therein, the authors retain all their rights. Acknowledgement Funding for the RFC Editor function is currently provided by the Internet Society. Further, the author Thomas Schierl of Fraunhofer HHI is sponsored by the European Commission under the contract number FP7-ICT-214063, project SEA. The authors want to thank Jonathan Lennox for his valuable comments on and input to the draft. 16. Open Issues 1) There is a list of remaining issues for decoding order recovery in session multiplexing, as documented in editing notes. 2) A lot of work needed for section9.2 (SDP parameters). 3) Bugs in RFC 3984 (see the BIS draft) need to be fixed also in the memo. 4) Clarify the usage of the new parameters like sprop-scalability- info, relation to SEI and usage in offer/answer. In the Design Team's conference call on 6 May 2008, it was decided that Ye-Kui would study and report. 5) Non-VCL NAL units, e.g. SEI messages and parameter sets, may be needed by an enhancement layer but not the base layer. However, according to SVC, within an access unit, these non-VCL NAL units must precede VCL NAL units in decoding order. In session multiplexing, should non-VCL NAL units be transported in the same session as the layer that requires the non-VCL NAL unit, or should they be always transported in the base session? It may be impossible to find out without parsing details which session respectively SPS/subset SPS a picture parameter set belongs to. It may make sense for simplicity to allow a MANE to include all of the non-VCL NAL units within all the sessions. In the Design Wenger, et al Expires December 30, 2008 [Page 80] Internet-Draft RTP Payload Format for SVC Video June 2008 Team's conference call on 6 May 2008, it was decided that Jonathan/Alex would provide text change for review, including handling of prefix NAL units. 6) sprop-spatial-resolution: do we need this for offering/answering spatial scalable layers? In this draft or a more generic draft? In the Design Team's conference call on 6 May 2008, it was decided that Ye-Kui will study and report. 17. Changes Log From draft-ietf-avt-rtp-svc-08 to draft-ietf-avt-rtp-svc-09 5-9 May 2008: YkW - Added Alex as an editor. Welcome! - Added text for session-multiplexing packetization modes - Updated section6.5 (Packetization Modes) and added section6.5.1 - Updated section6.6 (DON) and added section6.6.1 - Updated PACSI introductory text and semantics (section6.9) - Updated packetization rules for session multiplexing (section7.1) - Updated the de-packetization process for session multiplexing (section8.1) - Updated semantics of existing media type parameters and added new media type parameters ("pmode", "sprop-mst-interleave-depth", "sprop-remux-buf-req", "remux-buf-cap", "sprop-remux-init-buf- time", "sprop-mst-max-don-diff") in section9.1. - Removed obsolete comments. - Updated some definitions. - Updated one design principle regarding cases that must use RFC 3984 encapsulation. - Removed "(Informative)" from the title of section8 (De- Packetization Process) - same to be proposed to RFC 3984 bis. - Updated the open issues - Removed earlier changes log that can be found from earlier versions of the draft 13 May 2008: AE - Corrected definition of "highest RTP session" in "enhancement RTP session" definition. Wenger, et al Expires December 30, 2008 [Page 81] Internet-Draft RTP Payload Format for SVC Video June 2008 - Moved the triggering of the use of the S and E bits from the X flag to the Y flag, so that they are triggered together with TL0PICIDX. - Other minor editorial corrections for language. 13 May 2008: YkW - Removed some obsolete comments. 14 May 2008: AE - Moved definition of lower/higher/lowest/highest RTP session from definitions (part of "enhancement RTP session") to 8.1.1, right before they are first needed. The text assumes that base is lowest, and highest is the single RTP session which no other session depends on. - Modified definition of "operation point" to distinguish between the OP and the associated bitstream, as there may be multiple ways to construct it (i.e., removing or not removing unneeded NAL units below the operation point). - Renamed modes as follows: NITS->NI-T, NICD->NI-C, NICB->NI-TC, and AINT->I-C. This way the interleaving and DON process are evident from the acronym. - Rephrased "target NAL unit" definition (in Section 6.9) to be a bit more clear. 15 May 2008: YkW - Added a comment regarding the definition of operation point. - Updated open issues (removed the one on PACSI). - Changed the Word template used to generation of this I-D. From draft-ietf-avt-rtp-svc-09 to draft-ietf-avt-rtp-svc-10 30 May 2008: TS - Improved text in 7.1 for NI-T 02 June 2008: TS Wenger, et al Expires December 30, 2008 [Page 82] Internet-Draft RTP Payload Format for SVC Video June 2008 - Improved text on NI-T and NI-TC in section 8.1.1 - Added new packet type NI-MTAP in section 6.10 - Added informative text on placeholder NAL units to section 7.1.1 - Added text on placeholder NAL units to 8.1.1 - Changed text regarding the order of RTP packets are delivered from the RFC3550 process to the re-ordering process in section 8. 02 June 2008: JL - Added non-VCL and prefix NAL packetization rules sent by Jonathan Lennox in sections 7.1.4 and 7.1.5. - Added informative reordering algorithm. From draft-ietf-avt-rtp-svc-10 to draft-ietf-avt-rtp-svc-11 17 June 2008: TS - Addressed the following comments sent by Colin on 10 June 2008 to the mailing list: - Inserted table on NAL unit and packet types into section 3.3 - Corrected text on NAL unit header ext. in 3.3 - Corrected text in 3.4 - Corrected text in 5.1.1 on base layer VCL Nal unit types - Added reference to MANEs in section 6.1 - Used symbolic names for packetization modes - Corrected SDP examples - Removed H264 from media type section, TBD: text on "H264" needs to be moved - TBD: new name for session mux. - Improved text on NI-T mode From draft-ietf-avt-rtp-svc-11 to draft-ietf-avt-rtp-svc-12 30 June 2008: AE The entire text has been edited. The following offers highlights of the important changes. - Moved Section 3 (The SVC Codec) after "Scope" (Sec. 4) and "Definitions" (Sec. 5). Wenger, et al Expires December 30, 2008 [Page 83] Internet-Draft RTP Payload Format for SVC Video June 2008 - Changed section numbering style definitions in Word so that section references are properly formed (without a trailing "."). - Renumbered tables to run sequentially in the draft. - Removed [SVC] from the normative references. All references to [SVC] were converted to references to [H.264]. - Major rewrite of Introduction (and Abstract) to introduce concepts and facilitate understanding. The text from Scope has been moved there, and the Scope re-written to be inline with the intended content. - Definition of "session-multiplexing", is problematic. It states that in all cases there is a single RTP session, which is incorrect. The two cases are: 1) transmission on sepearate transport addresses, 2) transmission on a single transport address. In the former case there are clearly multiple RTP sessions. It is also not clear to me if the single transport address mode is supported in this memo or it should be explicitly disallowed. I put a comment to that effect in the text. - Session multiplexing: The text uses the term RTP base session and RTP enhancement session throughout, as well as RTP stream. Jonathan suggested multi-session stream, but that would create a lot of problems with the terminology in this draft (which says, e.g., that each RTP session carries an RTP stream). So I initially opted for the term "multi-session transmission". This is in-line with the RFC3550 definition of an "RTP session". We can easily search/replace to convert to "mutli-stream" transmission" if necessary (but it is not advisable, as the whole text needs to be reviewed for consistency). But due to the fact the draft's definition of "session multiplexing" explicitly says that it can be over the same transport or different transport addresses (thus using a single or multiple RTP sessions), I converged to "multi- source transmission" to indicate that the SVC stream is transmitted as streams coming from multiple sources. I think that nails it. Wenger, et al Expires December 30, 2008 [Page 84] Internet-Draft RTP Payload Format for SVC Video June 2008 - Defined acronyms SST (Single-Source Transmission) and MST (Multi- Source Transmission), and added in "Abbreviations". - Added definition for "NALU time" as "NAL unit effective timestamp". This is also inline with the concepts in 8.1.1. - Added ETS in the abbreviations for Effective Timestamp of a NAL unit. - It is noted that FU-A and FU-B are also carrying one NAL unit per RTP packet (since FU itself is defined as a NAL unit), so the statement that only in the single NAL unit mode you have one NAL unit per packet is, technically, incorrect. Since the intended use is present in the name of the mode (single NAL unit mode) I removed any dubious reference to this property. - Table 3.3 (now Table 4), changed PACSI from 'no' to 'yes' (in the non-interleaved column only). - Table 3.5 (now Table 6), changed PACSI from 'no' to 'yes'. But in 7.1.3 (I-C packetization rules) PACSIs are not allowed to have DONC. Why? - Moved Section 6.5 DON to the end of Section 6, as 6.10. Part of the reason is that the section requires that PACSI is already introduced. - "de-session-multiplexing buffer" was changed to "remultiplexing buffer", properly reflecting what's going on. - Changed all sprop-desemul-* to sprop-remux-*. - Changed all sprop-semul-* to sprop-mst-*. - Changed sprop-session-multiplexing-depth to sprop-mst-interleave- depth. - Added all acronyms encountered in the text in the "Abbreviations" section. Wenger, et al Expires December 30, 2008 [Page 85]