RFC 8680




Internet Engineering Task Force (IETF)                           V. Roca
Request for Comments: 8680                                         INRIA
Updates: 6363                                                   A. Begen
Category: Standards Track                                Networked Media
ISSN: 2070-1721                                             January 2020


  Forward Error Correction (FEC) Framework Extension to Sliding Window
                                 Codes

Abstract



   RFC 6363 describes a framework for using Forward Error Correction
   (FEC) codes to provide protection against packet loss.  The framework
   supports applying FEC to arbitrary packet flows over unreliable
   transport and is primarily intended for real-time, or streaming,
   media.  However, FECFRAME as per RFC 6363 is restricted to block FEC
   codes.  This document updates RFC 6363 to support FEC codes based on
   a sliding encoding window, in addition to block FEC codes, in a
   backward-compatible way.  During multicast/broadcast real-time
   content delivery, the use of sliding window codes significantly
   improves robustness in harsh environments, with less repair traffic
   and lower FEC-related added latency.

Status of This Memo



   This is an Internet Standards Track document.

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

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at
   https://www.rfc-editor.org/info/rfc8680.

Copyright Notice



   Copyright (c) 2020 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

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

Table of Contents



   1.  Introduction
   2.  Terminology
     2.1.  Definitions and Abbreviations
     2.2.  Requirements Language
   3.  Summary of Architecture Overview
   4.  Procedural Overview
     4.1.  General
     4.2.  Sender Operation with Sliding Window FEC Codes
     4.3.  Receiver Operation with Sliding Window FEC Codes
   5.  Protocol Specification
     5.1.  General
     5.2.  FEC Framework Configuration Information
     5.3.  FEC Scheme Requirements
   6.  Feedback
   7.  Transport Protocols
   8.  Congestion Control
   9.  Security Considerations
   10. Operations and Management Considerations
   11. IANA Considerations
   12. References
     12.1.  Normative References
     12.2.  Informative References
   Appendix A.  About Sliding Encoding Window Management
           (Informational)
   Acknowledgments

   Authors' Addresses



1.  Introduction



   Many applications need to transport a continuous stream of packetized
   data from a source (sender) to one or more destinations (receivers)
   over networks that do not provide guaranteed packet delivery.  In
   particular, packets may be lost, which is strictly the focus of this
   document: we assume that transmitted packets are either lost (e.g.,
   because of a congested router, a poor signal-to-noise ratio in a
   wireless network, or because the number of bit errors exceeds the
   correction capabilities of the physical-layer error-correcting code)
   or were received by the transport protocol without any corruption
   (i.e., the bit errors, if any, have been fixed by the physical-layer
   error-correcting code and therefore are hidden to the upper layers).

   For these use cases, Forward Error Correction (FEC) applied within
   the transport or application layer is an efficient technique to
   improve packet transmission robustness in the presence of packet
   losses (or "erasures") without going through packet retransmissions
   that create a delay often incompatible with real-time constraints.
   The FEC Building Block defined in [RFC5052] provides a framework for
   the definition of Content Delivery Protocols (CDPs) that make use of
   separately defined FEC schemes.  Any CDP defined according to the
   requirements of the FEC Building Block can then easily be used with
   any FEC scheme that is also defined according to the requirements of
   the FEC Building Block.

   Then, FECFRAME [RFC6363] provides a framework to define Content
   Delivery Protocols (CDPs) that provide FEC protection for arbitrary
   packet flows over an unreliable datagram service transport, such as
   UDP.  It is primarily intended for real-time or streaming media
   applications that are using broadcast, multicast, or on-demand
   delivery.  A subset of FECFRAME is currently part of the 3GPP Evolved
   Multimedia Broadcast/Multicast Service (eMBMS) standard [MBMSTS].

   However, [RFC6363] only considers block FEC schemes defined in
   accordance with the FEC Building Block [RFC5052] (e.g., [RFC6681],
   [RFC6816], or [RFC6865]).  These codes require the input flow(s) to
   be segmented into a sequence of blocks.  Then, FEC encoding (at a
   sender or an encoding middlebox) and decoding (at a receiver or a
   decoding middlebox) are both performed on a per-block basis.  For
   instance, if the current block encompasses the 100's to 119's source
   symbols (i.e., a block of size 20 symbols) of an input flow, encoding
   (and decoding) will be performed on this block independently of other
   blocks.  This approach has major impacts on FEC encoding and decoding
   delays.  The data packets of continuous media flow(s) may be passed
   to the transport layer immediately, without delay.  But the block
   creation time, which depends on the number of source symbols in this
   block, impacts both the FEC encoding delay (since encoding requires
   that all source symbols be known) and, mechanically, the packet loss
   recovery delay at a receiver (since no repair symbol for the current
   block can be generated and therefore received before that time).
   Therefore, a good value for the block size is necessarily a balance
   between the maximum FEC decoding latency at the receivers (which must
   be in line with the most stringent real-time requirement of the
   protected flow(s), hence an incentive to reduce the block size) and
   the desired robustness against long loss bursts (which increases with
   the block size, hence an incentive to increase this size).

   This document updates [RFC6363] in order to also support FEC codes
   based on a sliding encoding window (a.k.a., convolutional codes)
   [RFC8406].  This encoding window, either fixed or variable size,
   slides over the set of source symbols.  FEC encoding is launched
   whenever needed from the set of source symbols present in the sliding
   encoding window at that time.  This approach significantly reduces
   FEC-related latency, since repair symbols can be generated and passed
   to the transport layer on the fly at any time and can be regularly
   received by receivers to quickly recover packet losses.  Using
   sliding window FEC codes is therefore highly beneficial to real-time
   flows, one of the primary targets of FECFRAME.  [RFC8681] provides an
   example of such a FEC scheme for FECFRAME, which is built upon the
   simple sliding window Random Linear Code (RLC).

   This document is fully backward compatible with [RFC6363].  Indeed:

   *  This FECFRAME update does not prevent or compromise in any way the
      support of block FEC codes.  Both types of codes can nicely
      coexist, just like different block FEC schemes can coexist.

   *  Each sliding window FEC scheme is associated with a specific FEC
      Encoding ID subject to IANA registration, just like block FEC
      schemes.

   *  Any receiver -- for instance, a legacy receiver that only supports
      block FEC schemes -- can easily identify the FEC scheme used in a
      FECFRAME session.  Indeed, the FEC Encoding ID that identifies the
      FEC scheme is carried in FEC Framework Configuration Information
      (see Section 5.5 of [RFC6363]).  For instance, when the Session
      Description Protocol (SDP) is used to carry the FEC Framework
      Configuration Information, the FEC Encoding ID can be communicated
      in the "encoding-id=" parameter of a "fec-repair-flow" attribute
      [RFC6364].  This mechanism is the basic approach for a FECFRAME
      receiver to determine whether or not it supports the FEC scheme
      used in a given FECFRAME session.

   This document leverages on [RFC6363] and reuses its structure.  It
   proposes new sections specific to sliding window FEC codes whenever
   required.  The only exception is Section 3, which provides a quick
   summary of FECFRAME in order to facilitate the understanding of this
   document to readers not familiar with the concepts and terminology.

2.  Terminology



2.1.  Definitions and Abbreviations



   The following list of definitions and abbreviations is copied from
   [RFC6363], adding only the Block FEC Code, Sliding Window FEC Code,
   and Encoding/Decoding Window definitions (tagged with "ADDED"):

   Application Data Unit (ADU):
      The unit of source data provided as a payload to the transport
      layer.  For instance, it can be a payload containing the result of
      the RTP packetization of a compressed video frame.

   ADU Flow:
      A sequence of ADUs associated with a transport-layer flow
      identifier (such as the standard 5-tuple {source IP address,
      source port, destination IP address, destination port, transport
      protocol}).

   AL-FEC:
      Application-Layer Forward Error Correction.

   Application Protocol:
      Control protocol used to establish and control the source flow
      being protected, e.g., the Real-Time Streaming Protocol (RTSP).

   Content Delivery Protocol (CDP):
      A complete application protocol specification that, through the
      use of the framework defined in this document, is able to make use
      of FEC schemes to provide FEC capabilities.

   FEC Code:
      An algorithm for encoding data such that the encoded data flow is
      resilient to data loss.  Note that, in general, FEC codes may also
      be used to make a data flow resilient to corruption, but that is
      not considered in this document.

   Block FEC Code: (ADDED)
      A FEC code that operates on blocks, i.e., for which the input flow
      MUST be segmented into a sequence of blocks, with FEC encoding and
      decoding being performed independently on a per-block basis.

   Sliding Window FEC Code: (ADDED)
      A FEC code that can generate repair symbols on the fly, at any
      time, from the set of source symbols present in the sliding
      encoding window at that time.  These codes are also known as
      convolutional codes.

   FEC Framework:
      A protocol framework for the definition of Content Delivery
      Protocols using FEC, such as the framework defined in this
      document.

   FEC Framework Configuration Information:
      Information that controls the operation of the FEC Framework.

   FEC Payload ID:
      Information that identifies the contents and provides positional
      information of a packet with respect to the FEC scheme.

   FEC Repair Packet:
      At a sender (respectively, at a receiver), a payload submitted to
      (respectively, received from) the transport protocol containing
      one or more repair symbols along with a Repair FEC Payload ID and
      possibly an RTP header.

   FEC Scheme:
      A specification that defines the additional protocol aspects
      required to use a particular FEC code with the FEC Framework.

   FEC Source Packet:
      At a sender (respectively, at a receiver), a payload submitted to
      (respectively, received from) the transport protocol containing an
      ADU along with an optional Explicit Source FEC Payload ID.

   Repair Flow:
      The packet flow carrying FEC data.

   Repair FEC Payload ID:
      A FEC Payload ID specifically for use with repair packets.

   Source Flow:
      The packet flow to which FEC protection is to be applied.  A
      source flow consists of ADUs.

   Source FEC Payload ID:
      A FEC Payload ID specifically for use with source packets.

   Source Protocol:
      A protocol used for the source flow being protected, e.g., RTP.

   Transport Protocol:
      The protocol used for the transport of the source and repair
      flows.  This protocol needs to provide an unreliable datagram
      service, as UDP does ([RFC6363], Section 7).

   Encoding Window: (ADDED)
      Set of source symbols available at the sender/coding node that are
      used (with a Sliding Window FEC code) to generate a repair symbol.

   Decoding Window: (ADDED)
      Set of received or decoded source and repair symbols available at
      a receiver that are used (with a Sliding Window FEC code) to
      decode lost source symbols.

   Code Rate:
      The ratio between the number of source symbols and the number of
      encoding symbols.  By definition, the code rate is such that 0 <
      code rate <= 1.  A code rate close to 1 indicates that a small
      number of repair symbols have been produced during the encoding
      process.

   Encoding Symbol:
      Unit of data generated by the encoding process.  With systematic
      codes, source symbols are part of the encoding symbols.

   Packet Erasure Channel:
      A communication path where packets are either lost (e.g., in our
      case, by a congested router, or because the number of transmission
      errors exceeds the correction capabilities of the physical-layer
      code) or received.  When a packet is received, it is assumed that
      this packet is not corrupted (i.e., in our case, the bit errors,
      if any, are fixed by the physical-layer code and are therefore
      hidden to the upper layers).

   Repair Symbol:
      Encoding symbol that is not a source symbol.

   Source Block:
      Group of ADUs that are to be FEC protected as a single block.
      This notion is restricted to Block FEC codes.

   Source Symbol:
      Unit of data used during the encoding process.

   Systematic Code:
      FEC code in which the source symbols are part of the encoding
      symbols.

2.2.  Requirements Language



   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and
   "OPTIONAL" in this document are to be interpreted as described in
   BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all
   capitals, as shown here.

3.  Summary of Architecture Overview



   The architecture of Section 3 of [RFC6363] equally applies to this
   FECFRAME extension and is not repeated here.  However, this section
   includes a quick summary to facilitate the understanding of this
   document to readers not familiar with the concepts and terminology.

   +----------------------+
   |     Application      |
   +----------------------+
              |
              | (1) Application Data Units (ADUs)
              |
              v
   +----------------------+                           +----------------+
   |    FEC Framework     |                           |                |
   |                      |-------------------------->|   FEC Scheme   |
   |(2) Construct source  |(3) Source Block           |                |
   |    blocks            |                           |(4) FEC Encoding|
   |(6) Construct FEC     |<--------------------------|                |
   |    Source and Repair |                           |                |
   |    Packets           |(5) Explicit Source FEC    |                |
   +----------------------+    Payload IDs            +----------------+
              |                Repair FEC Payload IDs
              |                Repair symbols
              |
              |(7) FEC Source and Repair Packets
              v
   +----------------------+
   |  Transport Protocol  |
   +----------------------+

                Figure 1: FECFRAME Architecture at a Sender

   The FECFRAME architecture is illustrated in Figure 1 for a block FEC
   scheme from the sender's point of view.  It shows an application
   generating an ADU flow (other flows from other applications may
   coexist).  These ADUs of variable size must be somehow mapped to
   source symbols of a fixed size (this fixed size is a requirement of
   all FEC schemes, which comes from the way mathematical operations are
   applied to the symbols' content).  This is the goal of an ADU-to-
   symbols mapping process that is FEC scheme specific (see below).
   Once the source block is built, taking into account both the FEC
   scheme constraints (e.g., in terms of maximum source block size) and
   the application's flow constraints (e.g., in terms of real-time
   constraints), the associated source symbols are handed to the FEC
   scheme in order to produce an appropriate number of repair symbols.
   FEC Source Packets (containing ADUs) and FEC Repair Packets
   (containing one or more repair symbols each) are then generated and
   sent using an appropriate transport protocol (more precisely,
   Section 7 of [RFC6363] requires a transport protocol providing an
   unreliable datagram service, such as UDP).  In practice, FEC Source
   Packets may be passed to the transport layer as soon as available
   without having to wait for FEC encoding to take place.  In that case,
   a copy of the associated source symbols needs to be kept within
   FECFRAME for future FEC encoding purposes.

   At a receiver (not shown), FECFRAME processing operates in a similar
   way, taking as input the incoming FEC Source and Repair Packets
   received.  In case of FEC Source Packet losses, the FEC decoding of
   the associated block may recover all (in case of successful decoding)
   or a subset that is potentially empty (if decoding fails) of the
   missing source symbols.  After source-symbol-to-ADU mapping, when
   lost ADUs are recovered, they are then assigned to their respective
   flow (see below).  ADUs are returned to the application(s), either in
   their initial transmission order (in which case all ADUs received
   after a lost ADU will be delayed until FEC decoding has taken place)
   or not (in which case each ADU is returned as soon as it is received
   or recovered), depending on the application requirements.

   FECFRAME features two subtle mechanisms whose details are FEC scheme
   dependent:

   *  ADUs-to-source-symbols mapping: in order to manage variable size
      ADUs, FECFRAME and FEC schemes can use small, fixed-size symbols
      and create a mapping between ADUs and symbols.  The mapping
      details are FEC scheme dependent and must be defined in the
      associated document.  For instance, with certain FEC schemes, to
      each ADU, this mechanism prepends a length field (plus a flow
      identifier; see below) and pads the result to a multiple of the
      symbol size.  A small ADU may be mapped to a single source symbol,
      while a large one may be mapped to multiple symbols.

   *  Assignment of decoded ADUs to flows in multi-flow configurations:
      when multiple flows are multiplexed over the same FECFRAME
      instance, a problem is to assign a decoded ADU to the right flow
      (UDP port numbers and IP addresses traditionally used to map
      incoming ADUs to flows are not recovered during FEC decoding).
      The mapping details are FEC scheme dependent and must be defined
      in the associated document.  For instance, with certain FEC
      schemes, to make it possible, at the FECFRAME sending instance,
      each ADU is prepended with a flow identifier (1 byte) during the
      ADU-to-source-symbols mapping (see above).  The flow identifiers
      are also shared between all FECFRAME instances as part of the FEC
      Framework Configuration Information.  The ADU Information (ADUI),
      which includes the flow identifier, length, application payload,
      and padding, is then FEC protected.  Therefore, a decoded ADUI
      contains enough information to assign the ADU to the right flow.
      Note that a FEC scheme may also be restricted to the particular
      case of a single flow over a FECFRAME instance; that would make
      the above mechanism pointless.

   A few aspects are not covered by FECFRAME, namely:

   *  Section 8 of [RFC6363] does not detail any congestion control
      mechanisms and only provides high-level normative requirements.

   *  The possibility of having feedback from receiver(s) is considered
      out of scope, although such a mechanism may exist within the
      application (e.g., through RTP Control Protocol (RTCP) messages).

   *  Flow adaptation at a FECFRAME sender (e.g., how to set the FEC
      code rate based on transmission conditions) is not detailed, but
      it needs to comply with the congestion control normative
      requirements (see above).

4.  Procedural Overview



4.1.  General



   The general considerations of Section 4.1 of [RFC6363] that are
   specific to block FEC codes are not repeated here.

   With a Sliding Window FEC code, the FEC Source Packet MUST contain
   information to identify the position occupied by the ADU within the
   source flow in terms specific to the FEC scheme.  This information is
   known as the Source FEC Payload ID, and the FEC scheme is responsible
   for defining and interpreting it.

   With a Sliding Window FEC code, the FEC Repair Packets MUST contain
   information that identifies the relationship between the contained
   repair payloads and the original source symbols used during encoding.
   This information is known as the Repair FEC Payload ID, and the FEC
   scheme is responsible for defining and interpreting it.

   The sender operation ([RFC6363], Section 4.2) and receiver operation
   ([RFC6363], Section 4.3) are both specific to block FEC codes and are
   therefore omitted below.  The following two sections detail similar
   operations for Sliding Window FEC codes.

4.2.  Sender Operation with Sliding Window FEC Codes



   With a Sliding Window FEC scheme, the following operations,
   illustrated in Figure 2 for the generic case (non-RTP repair flows)
   and in Figure 3 for the case of RTP repair flows, describe a possible
   way to generate compliant source and repair flows:

   1.   A new ADU is provided by the application.

   2.   The FEC Framework communicates this ADU to the FEC scheme.

   3.   The sliding encoding window is updated by the FEC scheme.  The
        ADU-to-source-symbol mapping as well as the encoding window
        management details are both the responsibility of the FEC scheme
        and MUST be detailed there.  Appendix A provides non-normative
        hints about what FEC scheme designers need to consider.

   4.   The Source FEC Payload ID information of the source packet is
        determined by the FEC scheme.  If required by the FEC scheme,
        the Source FEC Payload ID is encoded into the Explicit Source
        FEC Payload ID field and returned to the FEC Framework.

   5.   The FEC Framework constructs the FEC Source Packet according to
        Figure 6 in [RFC6363], using the Explicit Source FEC Payload ID
        provided by the FEC scheme if applicable.

   6.   The FEC Source Packet is sent using normal transport-layer
        procedures.  This packet is sent using the same ADU flow
        identification information as would have been used for the
        original source packet if the FEC Framework were not present
        (e.g., the source and destination addresses and UDP port numbers
        on the IP datagram carrying the source packet will be the same
        whether or not the FEC Framework is applied).

   7.   When the FEC Framework needs to send one or several FEC Repair
        Packets (e.g., according to the target code rate), it asks the
        FEC scheme to create one or several repair packet payloads from
        the current sliding encoding window along with their Repair FEC
        Payload ID.

   8.   The Repair FEC Payload IDs and repair packet payloads are
        provided back by the FEC scheme to the FEC Framework.



   9.   The FEC Framework constructs FEC Repair Packets according to
        Figure 7 in [RFC6363], using the FEC Payload IDs and repair
        packet payloads provided by the FEC scheme.

   10.  The FEC Repair Packets are sent using normal transport-layer
        procedures.  The port(s) and multicast group(s) to be used for
        FEC Repair Packets are defined in the FEC Framework
        Configuration Information.

   +----------------------+
   |     Application      |
   +----------------------+
              |
              | (1) New Application Data Unit (ADU)
              v
   +---------------------+                           +----------------+
   |    FEC Framework    |                           |   FEC Scheme   |
   |                     |-------------------------->|                |
   |                     | (2) New ADU               |(3) Update of   |
   |                     |                           |    encoding    |
   |                     |<--------------------------|    window      |
   |(5) Construct FEC    | (4) Explicit Source       |                |
   |    Source Packet    |     FEC Payload ID(s)     |(7) FEC         |
   |                     |<--------------------------|    encoding    |
   |(9) Construct FEC    | (8) Repair FEC Payload ID |                |
   |    Repair Packet(s) |     + Repair symbol(s)    +----------------+
   +---------------------+
              |
              | (6)  FEC Source Packet
              | (10) FEC Repair Packets
              v
   +----------------------+
   |  Transport Protocol  |
   +----------------------+

          Figure 2: Sender Operation with Sliding Window FEC Codes

   +----------------------+
   |     Application      |
   +----------------------+
              |
              | (1) New Application Data Unit (ADU)
              v
   +---------------------+                           +----------------+
   |    FEC Framework    |                           |   FEC Scheme   |
   |                     |-------------------------->|                |
   |                     | (2) New ADU               |(3) Update of   |
   |                     |                           |    encoding    |
   |                     |<--------------------------|    window      |
   |(5) Construct FEC    | (4) Explicit Source       |                |
   |    Source Packet    |     FEC Payload ID(s)     |(7) FEC         |
   |                     |<--------------------------|    encoding    |
   |(9) Construct FEC    | (8) Repair FEC Payload ID |                |
   |    Repair Packet(s) |     + Repair symbol(s)    +----------------+
   +---------------------+
       |             |
       |(6) Source   |(10) Repair payloads
       |    packets  |
       |      + -- -- -- -- -+
       |      |     RTP      |
       |      +-- -- -- -- --+
       v             v
   +----------------------+
   |  Transport Protocol  |
   +----------------------+

      Figure 3: Sender Operation with Sliding Window FEC Codes and RTP
                                Repair Flows

4.3.  Receiver Operation with Sliding Window FEC Codes



   With a Sliding Window FEC scheme, the following operations are
   illustrated in Figure 4 for the generic case (non-RTP repair flows)
   and in Figure 5 for the case of RTP repair flows.  The only
   differences with respect to block FEC codes lie in steps (4) and (5).
   Therefore, this section does not repeat the other steps of
   Section 4.3 of [RFC6363] ("Receiver Operation").  The new steps (4)
   and (5) are:

   4.  The FEC scheme uses the received FEC Payload IDs (and derived FEC
       Source Payload IDs when the Explicit Source FEC Payload ID field
       is not used) to insert source and repair packets into the
       decoding window in the right way.  If at least one source packet
       is missing and at least one repair packet has been received, then
       FEC decoding is attempted to recover the missing source payloads.
       The FEC scheme determines whether source packets have been lost
       and whether enough repair packets have been received to decode
       any or all of the missing source payloads.

   5.  The FEC scheme returns the received and decoded ADUs to the FEC
       Framework, along with indications of any ADUs that were missing
       and could not be decoded.

   +----------------------+
   |     Application      |
   +----------------------+
              ^
              |(6) ADUs
              |
   +----------------------+                           +----------------+
   |    FEC Framework     |                           |   FEC Scheme   |
   |                      |<--------------------------|                |
   |(2)Extract FEC Payload|(5) ADUs                   |(4) FEC Decoding|
   |   IDs and pass IDs & |-------------------------->|                |
   |   payloads to FEC    |(3) Explicit Source FEC    +----------------+
   |   scheme             |            Payload IDs
   +----------------------+    Repair FEC Payload IDs
              ^                Source payloads
              |                Repair payloads
              |(1) FEC Source
              |    and Repair Packets
   +----------------------+
   |  Transport Protocol  |
   +----------------------+

         Figure 4: Receiver Operation with Sliding Window FEC Codes

   +----------------------+
   |     Application      |
   +----------------------+
              ^
              |(6) ADUs
              |
   +----------------------+                           +----------------+
   |    FEC Framework     |                           |   FEC Scheme   |
   |                      |<--------------------------|                |
   |(2)Extract FEC Payload|(5) ADUs                   |(4) FEC Decoding|
   |   IDs and pass IDs & |-------------------------->|                |
   |   payloads to FEC    |(3) Explicit Source FEC    +----------------+
   |   scheme             |            Payload IDs
   +----------------------+    Repair FEC Payload IDs
       ^             ^         Source payloads
       |             |         Repair payloads
       |Source pkts  |Repair payloads
       |             |
   +-- |- -- -- -- -- -- -+
   |RTP| | RTP Processing |
   |   | +-- -- -- --|-- -+
   | +-- -- -- -- -- |--+ |
   | | RTP Demux        | |
   +-- -- -- -- -- -- -- -+
              ^
              |(1) FEC Source and Repair Packets
              |
   +----------------------+
   |  Transport Protocol  |
   +----------------------+

       Figure 5: Receiver Operation with Sliding Window FEC Codes and
                              RTP Repair Flows

5.  Protocol Specification



5.1.  General



   This section discusses the protocol elements for the FEC Framework
   specific to Sliding Window FEC schemes.  The global formats of source
   data packets (i.e., [RFC6363], Figure 6) and repair data packets
   (i.e., [RFC6363], Figures 7 and 8) remain the same with Sliding
   Window FEC codes.  They are not repeated here.

5.2.  FEC Framework Configuration Information



   The FEC Framework Configuration Information considerations of
   Section 5.5 of [RFC6363] equally apply to this FECFRAME extension and
   are not repeated here.

5.3.  FEC Scheme Requirements



   The FEC scheme requirements of Section 5.6 of [RFC6363] mostly apply
   to this FECFRAME extension and are not repeated here.  An exception,
   though, is the "full specification of the FEC code", item (4), which
   is specific to block FEC codes.  In case of a Sliding Window FEC
   scheme, then the following item (4-bis) applies:

   4-bis.
       A full specification of the Sliding Window FEC code.

       This specification MUST precisely define the valid FEC-Scheme-
       Specific Information values, the valid FEC Payload ID values, and
       the valid packet payload sizes (where "packet payload" refers to
       the space within a packet dedicated to carrying encoding
       symbols).

       Furthermore, given valid values of the FEC-Scheme-Specific
       Information, a valid Repair FEC Payload ID value, a valid packet
       payload size, and a valid encoding window (i.e., a set of source
       symbols), the specification MUST uniquely define the values of
       the encoding symbol (or symbols) to be included in the repair
       packet payload with the given Repair FEC Payload ID value.

   Additionally, the FEC scheme associated with a Sliding Window FEC
   code:

   *  MUST define the relationships between ADUs and the associated
      source symbols (mapping).

   *  MUST define the management of the encoding window that slides over
      the set of ADUs.  Appendix A provides non-normative hints about
      what FEC scheme designers need to consider.

   *  MUST define the management of the decoding window.  This usually
      consists of managing a system of linear equations (for a linear
      FEC code).

6.  Feedback



   The discussion in Section 6 of [RFC6363] equally applies to this
   FECFRAME extension and is not repeated here.

7.  Transport Protocols



   The discussion in Section 7 of [RFC6363] equally applies to this
   FECFRAME extension and is not repeated here.

8.  Congestion Control

   The discussion in Section 8 of [RFC6363] equally applies to this
   FECFRAME extension and is not repeated here.

9.  Security Considerations



   This FECFRAME extension does not add any new security considerations.
   All the considerations of Section 9 of [RFC6363] apply to this
   document as well.  However, for the sake of completeness, the
   following goal can be added to the list provided in Section 9.1 of
   [RFC6363] ("Problem Statement"):

   *  Attacks can try to corrupt source flows in order to modify the
      receiver application's behavior (as opposed to just denying
      service).

10.  Operations and Management Considerations



   This FECFRAME extension does not add any new Operations and
   Management Considerations.  All the considerations of Section 10 of
   [RFC6363] apply to this document as well.

11.  IANA Considerations



   This document has no IANA actions.

   A FEC scheme for use with this FEC Framework is identified via its
   FEC Encoding ID.  It is subject to IANA registration in the "FEC
   Framework (FECFRAME) FEC Encoding IDs" registry.  All the rules of
   Section 11 of [RFC6363] apply and are not repeated here.

12.  References



12.1.  Normative References



   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119,
              DOI 10.17487/RFC2119, March 1997,
              <https://www.rfc-editor.org/info/rfc2119>.

   [RFC6363]  Watson, M., Begen, A., and V. Roca, "Forward Error
              Correction (FEC) Framework", RFC 6363,
              DOI 10.17487/RFC6363, October 2011,
              <https://www.rfc-editor.org/info/rfc6363>.

   [RFC8174]  Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC
              2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174,
              May 2017, <https://www.rfc-editor.org/info/rfc8174>.

12.2.  Informative References



   [MBMSTS]   3GPP, "Multimedia Broadcast/Multicast Service (MBMS);
              Protocols and codecs", 3GPP TS 26.346, March 2009,
              <http://ftp.3gpp.org/specs/html-info/26346.htm>.

   [RFC5052]  Watson, M., Luby, M., and L. Vicisano, "Forward Error
              Correction (FEC) Building Block", RFC 5052,
              DOI 10.17487/RFC5052, August 2007,
              <https://www.rfc-editor.org/info/rfc5052>.

   [RFC6364]  Begen, A., "Session Description Protocol Elements for the
              Forward Error Correction (FEC) Framework", RFC 6364,
              DOI 10.17487/RFC6364, October 2011,
              <https://www.rfc-editor.org/info/rfc6364>.

   [RFC6681]  Watson, M., Stockhammer, T., and M. Luby, "Raptor Forward
              Error Correction (FEC) Schemes for FECFRAME", RFC 6681,
              DOI 10.17487/RFC6681, August 2012,
              <https://www.rfc-editor.org/info/rfc6681>.

   [RFC6816]  Roca, V., Cunche, M., and J. Lacan, "Simple Low-Density
              Parity Check (LDPC) Staircase Forward Error Correction
              (FEC) Scheme for FECFRAME", RFC 6816,
              DOI 10.17487/RFC6816, December 2012,
              <https://www.rfc-editor.org/info/rfc6816>.

   [RFC6865]  Roca, V., Cunche, M., Lacan, J., Bouabdallah, A., and K.
              Matsuzono, "Simple Reed-Solomon Forward Error Correction
              (FEC) Scheme for FECFRAME", RFC 6865,
              DOI 10.17487/RFC6865, February 2013,
              <https://www.rfc-editor.org/info/rfc6865>.

   [RFC8406]  Adamson, B., Adjih, C., Bilbao, J., Firoiu, V., Fitzek,
              F., Ghanem, S., Lochin, E., Masucci, A., Montpetit, M-J.,
              Pedersen, M., Peralta, G., Roca, V., Ed., Saxena, P., and
              S. Sivakumar, "Taxonomy of Coding Techniques for Efficient
              Network Communications", RFC 8406, DOI 10.17487/RFC8406,
              June 2018, <https://www.rfc-editor.org/info/rfc8406>.

   [RFC8681]  Roca, V. and B. Teibi, "Sliding Window Random Linear Code
              (RLC) Forward Erasure Correction (FEC) Schemes for
              FECFRAME", RFC 8681, DOI 10.17487/RFC8681, January 2020,
              <https://www.rfc-editor.org/info/rfc8681>.

Appendix A.  About Sliding Encoding Window Management (Informational)



   The FEC Framework does not specify the management of the sliding
   encoding window, which is the responsibility of the FEC scheme.  This
   annex only provides a few informational hints.

   Source symbols are added to the sliding encoding window each time a
   new ADU is available at the sender after the ADU-to-source-symbol
   mapping specific to the FEC scheme has been done.

   Source symbols are removed from the sliding encoding window.  For
   instance:

   *  After a certain delay, when an "old" ADU of a real-time flow times
      out.  The source symbol retention delay in the sliding encoding
      window should therefore be initialized according to the real-time
      features of incoming flow(s) when applicable.

   *  Once the sliding encoding window has reached its maximum size
      (there is usually an upper limit to the sliding encoding window
      size).  In that case, the oldest symbol is removed each time a new
      source symbol is added.

   Several considerations can impact the management of this sliding
   encoding window:

   *  At the source flows level: real-time constraints can limit the
      total time during which source symbols can remain in the encoding
      window.

   *  At the FEC code level: theoretical or practical limitations (e.g.,
      because of computational complexity) can limit the number of
      source symbols in the encoding window.

   *  At the FEC scheme level: signaling and window management are
      intrinsically related.  For instance, an encoding window composed
      of a nonsequential set of source symbols requires appropriate
      signaling to inform a receiver of the composition of the encoding
      window, and the associated transmission overhead can limit the
      maximum encoding window size.  On the contrary, an encoding window
      always composed of a sequential set of source symbols simplifies
      signaling: providing the identity of the first source symbol plus
      its number is sufficient, which creates a fixed and relatively
      small transmission overhead.

Acknowledgments

   The authors would like to thank Christer Holmberg, David Black, Gorry
   Fairhurst, Emmanuel Lochin, Spencer Dawkins, Ben Campbell, Benjamin
   Kaduk, Eric Rescorla, Adam Roach, and Greg Skinner for their valuable
   feedback on this document.  This document being an extension of
   [RFC6363], the authors would also like to thank Mark Watson as the
   main author of that RFC.

Authors' Addresses



   Vincent Roca
   INRIA
   Univ. Grenoble Alpes
   France

   Email: vincent.roca@inria.fr


   Ali Begen
   Networked Media
   Konya/
   Turkey

   Email: ali.begen@networked.media