RFC 2733
This document is obsolete. Please refer to RFC 5109.

Network Working Group                                       J. Rosenberg
Request for Comments: 2733                                   dynamicsoft
Category: Standards Track                                 H. Schulzrinne
                                                     Columbia University
                                                           December 1999

       An RTP Payload Format for Generic Forward Error Correction

Status of this Memo

   This document specifies an Internet standards track protocol for the
   Internet community, and requests discussion and suggestions for
   improvements.  Please refer to the current edition of the "Internet
   Official Protocol Standards" (STD 1) for the standardization state
   and status of this protocol.  Distribution of this memo is unlimited.

Copyright Notice

   Copyright (C) The Internet Society (1999).  All Rights Reserved.


   This document specifies a payload format for generic forward error
   correction of media encapsulated in RTP. It is engineered for FEC
   algorithms based on the exclusive-or (parity) operation. The payload
   format allows end systems to transmit using arbitrary block lengths
   and parity schemes. It also allows for the recovery of both the
   payload and critical RTP header fields. Since FEC is sent as a
   separate stream, it is backwards compatible with non-FEC capable
   hosts, so that receivers which do not wish to implement FEC can just
   ignore the extensions.

Table of Contents

   1     Introduction ...........................................    2
   2     Terminology ............................................    2
   3     Basic Operation ........................................    3
   4     Parity Codes ...........................................    5
   5     RTP Media Packet Structure .............................    6
   6     FEC Packet Structure ...................................    7
   6.1   RTP Header of FEC Packets ..............................    7
   6.2   FEC Header .............................................    7
   7     Protection Operation ...................................    9
   8     Recovery Procedures ....................................   10
   8.1   Reconstruction .........................................   10
   8.2   Determination of When to Recover .......................   12

Rosenberg & Schulzrinne     Standards Track                     [Page 1]

RFC 2733                      Generic FEC                  December 1999

   9     Example ................................................   16
   10    Use with Redundant Encodings ...........................   17
   11    Indicating FEC Usage in SDP ............................   20
   11.1  FEC as a Separate Stream ...............................   20
   11.2  Use with Redundant Encodings ...........................   21
   11.3  Usage with RTSP ........................................   22
   12    Security Considerations ................................   23
   13    Acknowledgments ........................................   24
   14    Authors' Addresses .....................................   24
   15    Bibliography ...........................................   25
   16    Full Copyright Statement ...............................   26

1 Introduction

   The quality of packet voice on the Internet has been mediocre due, in
   part, to high packet loss rates. This is especially true on wide-area
   connections. Unfortunately, the strict delay requirements of real-
   time multimedia usually eliminate the possibility of retransmissions.

   It is for this reason that forward error correction (FEC) has been
   proposed to compensate for packet loss in the Internet [1] [2]. In
   particular, the use of traditional error correcting codes, such as
   parity, Reed-Solomon, and Hamming codes, has attracted attention. To
   support these mechanisms, protocol support is required.

   This document defines a payload format for RTP [3] which allows for
   generic forward error correction of real time media. In this context,
   generic means that the FEC protocol is (1) independent of the nature
   of the media being protected, be it audio, video, or otherwise, (2)
   flexible enough to support a wide variety of FEC mechanisms, (3)
   designed for adaptivity so that the FEC technique can be modified
   easily without out of band signaling, and (4) supportive of a number
   of different mechanisms for transporting the FEC packets.

2 Terminology

   The following terms are used throughout this document:

       Media Payload: is a piece of raw, un-protected user data which
            is to be transmitted from the sender. The media payload is
            placed inside of an RTP packet.

       Media Header: is the RTP header for the packet containing the
            media payload.

       Media Packet: The combination of a media payload and media
            header is called a media packet.

Rosenberg & Schulzrinne     Standards Track                     [Page 2]

RFC 2733                      Generic FEC                  December 1999

       FEC Packet: The forward error correction algorithms at the
            transmitter take the media packets as an input. They output
            both the media packets that they are passed, and new
            packets called FEC packets. The FEC packets are formatted
            according to the rules specified in this document.

       FEC Header: The FEC header is the header information contained
            in an FEC packet.

       FEC Payload: The FEC payload is the payload in an FEC packet.

       Associated: An FEC packet is said to be "associated" with one or
            more media packets when those media packets are used to
            generate the FEC packet (by use of the exclusive or

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   document are to be interpreted as described in RFC 2119 [4].

3 Basic Operation

   The payload format described here is used whenever a participant in
   an RTP session would like to protect a media stream it is sending
   with forward error correction (FEC). The FEC supported by the format
   are those codes based on simple exclusive or (xor) parities. The
   sender takes some set of packets from the media stream, and applies
   an xor operation across the payloads. The sender also applies the xor
   operation over components of the RTP headers. Based on the procedures
   defined here, the result is an RTP packet containing FEC information.
   This packet can be used at the receiver to recover any one of the
   packets used to generate the FEC packet. This document does not
   mandate the particular set of media packets combined to generate an
   FEC packet (such a set [is] referred to as a code). Use of differing
   sets results in a tradeoff between overhead, delay, and
   recoverability.  Section 4 outlines some possible combinations.

   The payload format contains information that allows the sender to
   tell the receiver exactly which media packets have been used to
   generate the FEC. Specifically, each FEC packet contains a bitmask,
   called the offset mask, containing 24 bits. If bit i in the mask is
   set to 1, the media packet with sequence number N + i was used to
   generate this FEC packet. N is called the sequence number base, and
   is sent in the FEC packet as well. The offset mask and payload type
   are sufficient to signal arbitrary parity based forward error
   correction schemes with little overhead.

Rosenberg & Schulzrinne     Standards Track                     [Page 3]

RFC 2733                      Generic FEC                  December 1999

   This document also describes procedures that allow the receiver to
   make use of the FEC without having to know the details of specific
   codes. This allows the sender much flexibility; it can adapt the code
   in use based on network conditions, and be certain the receivers can
   still make use of the FEC for recovery.

   As the sender generates FEC packets, they are sent to the receivers.
   The sender still usually sends the original media stream, as if there
   were no FEC. This allows the media stream to still be used by
   receivers who are not FEC capable. However, some FEC codes do not
   require the original media to be sent; the FEC stream is sufficient
   for recovery. These codes have the drawback that all receivers must
   be FEC capable. However, they are supported by this format.

   The FEC packets are not sent in the same RTP stream as the media
   packets. They can be sent as a separate stream, or as a secondary
   codec in the redundant codec payload format [5]. When sent as a
   separate stream, the FEC packets have their own sequence number
   space. Although the timestamps for the FEC packets are derived from
   the media packets, they increment monotonically. FEC packet streams
   thus work well with any header compression mechanism which requires
   fixed deltas between fields in the packet header.

   This document does not prescribe the definition of "separate
   streams", but leaves this to applications and higher level protocols
   to define. For multicast, the separate stream may be implemented by
   separate multicast groups, different ports in the same group, or by a
   different SSRC within the same group/port. For unicast, different
   ports or different SSRC may be used. Each of these approaches has
   drawbacks and benefits which depend on the application.

   At the receiver, the FEC and original media are received. If no media
   packets are lost, the FEC can be ignored. In the event of loss, the
   FEC packets can be combined with other media and FEC packets that
   have been received, resulting in recovery of missing media packets.
   The recovery is exact; the payload is perfectly reconstructed, along
   with most components of the header.

   RTP packets which contain data formatted according to this
   specification (i.e., FEC packets) are signaled using dynamic RTP
   payload types.

Rosenberg & Schulzrinne     Standards Track                     [Page 4]

RFC 2733                      Generic FEC                  December 1999

4 Parity Codes

   For brevity, we define the function f(x,y,..) to be the XOR (parity)
   operator applied to the packets x,y,... The output of this function
   is another packet, called the parity packet. For simplicity, we
   assume here that the parity packet is computed as the bitwise XOR of
   the input packets. The exact procedure is specified in section 6.

   Recovery of data packets using parity codes is accomplished by
   generating one or more parity packets over a group of data packets.
   To be effective, the parity packets must be generated by linearly
   independent combinations of data packets. The particular combination
   is called a parity code. One class of codes takes a group of k data
   packets, and generates n-k parity packets. There are a large number
   of possible parity codes for a given n,k. The payload format does not
   mandate a particular code.

   For example, consider a parity code which generates a single parity
   packet over two data packets. If the original media packets are
   a,b,c,d, the packets generated by the sender are:

   a        b        c        d               <-- media stream
              f(a,b)            f(c,d)        <-- FEC stream

   where time increases to the right. In this example, the error
   correction scheme (we use the terms scheme and code interchangeably)
   introduces a 50% overhead. But if b is lost, a and f(a,b) can be used
   to recover b.

   Some additional codes are listed below. In each, the original media
   stream consists of packets a,b,c,d and so on.

   Scheme 1

   This scheme is the similar to the one in the example above. However,
   instead of sending b, followed by f(a,b), f(a,b) is sent before b.
   Doing this clearly requires additional delay at the sender. However,
   if allows some bursts of two consecutive packet losses to be
   recovered. The packets generated by the sender look like:

   a        b        c        d        e        <-- media stream
     f(a,b)   f(b,c)   f(c,d)   f(d,e)          <-- FEC stream

Rosenberg & Schulzrinne     Standards Track                     [Page 5]

RFC 2733                      Generic FEC                  December 1999

   Scheme 2

   It is not strictly necessary for the original media stream to be
   transmitted. In this scheme, only FEC packets are transmitted.  This
   scheme allows for recovery of all single packet losses and some
   consecutive packet losses, but with slightly less overhead than
   scheme 1. The packets generated by the sender look like:

   f(a,b)  f(a,c)  f(a,b,c)  f(c,d)  f(c,e)  f(c,d,e)  <-- FEC stream

   Scheme 3

   This scheme requires the receiver to wait an additional four packet
   intervals to recover the original media packets. However, it can
   recover from one, two or three consecutive packet losses. The packets
   generated by the sender look like:

   a         b          c                    d     <-- media stream
               f(a,b,c)    f(a,c,d) f(a,b,d)       <-- FEC stream

5 RTP Media Packet Structure

   The formatting of the media packets is unaffected by FEC. If the FEC
   is sent as a separate stream, the media packets are sent as if there
   was no FEC. If the FEC is being sent as a redundant codec, the media
   packets are sent as the main codec as defined in RFC 2198 [5].

   This lends to a very efficient encoding. When little (or no) FEC is
   used, there are mostly media packets being sent. This means that the
   overhead (present in FEC packets only) tracks the amount of FEC in

Rosenberg & Schulzrinne     Standards Track                     [Page 6]

RFC 2733                      Generic FEC                  December 1999

6 FEC Packet Structure

   An FEC packet is constructed by placing an FEC header and FEC payload
   in the RTP payload, as shown in Figure 1:

   |                         RTP Header                            |
   |                         FEC Header                            |
   |                         FEC Payload                           |
   |                                                               |

   Figure 1: FEC Packet Structure

6.1 RTP Header of FEC Packets

   The version field is set to 2. The padding bit is computed via the
   protection operation, defined below. The extension bit is also
   computed via the protection operation. The SSRC value will generally
   be the same as the SSRC value of the media stream it protects. It MAY
   be different if the FEC stream is being demultiplexed via the SSRC
   value. The CC value is computed via the protection operation. The
   CSRC list is never present, independent of the value of the CC field.
   The extension is never present, independent of the value of the X
   bit. The marker bit is computed via the protection operation.

   The sequence number has the standard definition: it MUST be one
   higher than the sequence number in the previously transmitted FEC
   packet. The timestamp MUST be set to the value of the media RTP clock
   at the instant the FEC packet is transmitted. This results in the TS
   value in FEC packets to be monotonically increasing, independent of
   the FEC scheme.

   The payload type for the FEC packet is determined through dynamic,
   out of band means. According to RFC 1889 [3], RTP participants which
   cannot recognize a payload type must discard it. This provides
   backwards compatibility. The FEC mechanisms can then be used in a
   multicast group with mixed FEC-capable and FEC-incapable receivers.

6.2 FEC Header

   This header is 12 bytes. The format of the header is shown in Figure
   2, and consists of an SN base field, length recovery field, E field,
   PT recovery field, mask field and TS recovery field.

Rosenberg & Schulzrinne     Standards Track                     [Page 7]

RFC 2733                      Generic FEC                  December 1999

   |      SN base                  |        length recovery        |
   |E| PT recovery |                 mask                          |
   |                          TS recovery                          |

   Figure 2: Parity Header Format

   The length recovery field is used to determine the length of any
   recovered packets. It is computed via the protection operation
   applied to the unsigned network-ordered 16 bit representation of the
   sums of the lengths (in bytes) of the media payload, CSRC list,
   extension and padding of media packets associated with this FEC
   packet (in other words, the CSRC list, extension, and padding, if
   present, are "counted" as part of the payload). This allows the FEC
   procedure to be applied even when the lengths of the media packets
   are not identical. For example, assume an FEC packet is being
   generated by xor'ing two media packets together. The length of the
   two media packets are 3 (0b011) and 5 (0b101) bytes, respectively.
   The length recovery field is then encoded as 0b011 xor 0b101 = 0b110.

   The E bit indicates a header extension. Implementations conforming to
   this version of the specification MUST set this bit to zero.

   The PT recovery field is obtained via the protection operation
   applied to the payload type values of the media packets associated
   with the FEC packet.

   The mask field is 24 bits. If bit i in the mask is set to 1, then the
   media packet with sequence number N + i is associated with this FEC
   packet, where N is the SN Base field in the FEC packet header. The
   least significant bit corresponds to i=0, and the most significant to

   The SN base field MUST be set to the minimum sequence number of those
   media packets protected by FEC. This allows for the FEC operation to
   extend over any string of at most 24 packets.

   The TS recovery field is computed via the protection operation
   applied to the timestamps of the media packets associated with this
   FEC packet. This allows the timestamp to be completely recovered.

   The payload of the FEC packet is the protection operation applied to
   the concatenation of the CSRC list, RTP extension, media payload, and
   padding of the media packets associated with the FEC packet.

Rosenberg & Schulzrinne     Standards Track                     [Page 8]

RFC 2733                      Generic FEC                  December 1999

   Note that it's possible for the FEC packet to be slightly larger than
   the media packets it protects (due to the presence of the FEC
   header). This could cause difficulties if this results in the FEC
   packet exceeding the Maximum Transmission Unit size for the path
   along which it is sent.

7 Protection Operation

   The protection operation involves concatenating specific fields from
   the RTP header of the media packet, appending the payload, padding
   with zeroes, and then computing the xor across the resulting bit
   strings. The resulting bit string is used to generate the FEC packet.

   The following procedure MAY be followed for the protection operation.
   Other procedures MAY be followed, but the end result MUST be
   identical to the one described here. For each media packet to be
   protected, a bit string is generated by concatenating the following
   fields together in the order specifed:

      o Padding Bit (1 bit)

      o Extension Bit (1 bit)

      o CC bits (4 bits)

      o Marker bit (1 bit)

      o Payload Type (7 bits)

      o Timestamp (32 bits)

      o Unsigned network-ordered 16 bit representation of the sum of
        the lengths (in bytes) of the CSRC List, length of the padding,
        length of the extension, and length of the media payload (16

      o if CC is nonzero, the CSRC List (variable length)

      o if X is 1, the Header Extension (variable length)

      o the payload (variable length)

      o Padding, if present (variable length)

   Note that the Padding Bit (first entry above) forms the most
   significant bit of the bit string.

Rosenberg & Schulzrinne     Standards Track                     [Page 9]

RFC 2733                      Generic FEC                  December 1999

   If the lengths of the bit strings are not equal, each bit string that
   is shorter than the length of the longest, MUST be padded to the
   length of the longest. Any value for the pad may be used. The pad
   MUST be added at the end of the bit string.

   The parity operation is then applied across the bit strings. The
   result is the bit string used to build the FEC packet. Call this the
   FEC bit string.

   The first (most significant) bit in the FEC bit string is written
   into the Padding Bit of the FEC packet. The second bit in the FEC bit
   string is written into the Extension bit of the FEC packet. The next
   four bits of the FEC bit string are written into the CC field of the
   FEC packet. The next bit of the FEC bit string is written into the
   marker bit of the FEC packet. The next 7 bits of the FEC bit string
   are written into the PT recovery field in the FEC packet header. The
   next 32 bits of the FEC bit string are written into the TS recovery
   field in the packet header. The next 16 bits are written into the
   length recovery field in the FEC packet header. The remaining bits
   are set to be the payload of the FEC packet.

8 Recovery Procedures

   The FEC packets allow end systems to recover from the loss of media
   packets. All of the header fields of the missing packets, including
   CSRC lists, extensions, padding bits, marker and payload type, are
   recoverable.  This section describes the procedure for performing
   this recovery.

   Recovery requires two distinct operations. The first determines which
   packets (media and FEC) must be combined in order to recover a
   missing packet. Once this is done, the second step is to actually
   reconstruct the data. The second step MUST be performed as described
   below. The first step MAY be based on any algorithm chosen by the
   implementer. Different algorithms result in a tradeoff between
   complexity and the ability to recover missing packets if at all

8.1 Reconstruction

   Let T be the list of packets (FEC and media) which can be combined to
   recover some media packet xi. The procedure is as follows:

       1.   For the media packets in T, compute the bit string as
            described in the protection operation of the previous

Rosenberg & Schulzrinne     Standards Track                    [Page 10]

RFC 2733                      Generic FEC                  December 1999

       2.   For the FEC packet in T, compute the bit string in the same
            fashion, except use the PT Recovery instead of Payload Type,
            TS Recovery instead of Timestamp,  and always set the CSRC
            list, extension, and padding to null.

       3.   If any of the bit strings generated from the media packets
            are shorter than the bit string generated from the FEC
            packet, pad them to be the same length as the bit string
            generated from the FEC. The padding MUST be added at the
            end of the bit string, and MAY be of any value.

       4.   Perform the exclusive or (parity) operation across the bit
            strings, resulting in a recovery bit string.

       5.   Create a new packet with the standard 12 byte RTP header
            and no payload.

       6.   Set the version of the new packet to 2.

       7.   Set the Padding bit in the new packet to the first bit in
            the recovery bit string.

       8.   Set the Extension bit in the new packet to the second bit
            in the recovery bit string.

       9.   Set the CC field to the next four bits in the recovery bit

       10.  Set the marker bit in the new packet to the next bit in the
            recovery bit string.

       11.  Set the payload type in the new packet to the next 7 bits
            in the recovery bit string.

       12.  Set the SN field in the new packet to xi.

       13.  Set the TS field in the new packet to the next 32 bits in
            the recovery bit string.

       14.  Take the next 16 bits of the recovery bit string. Whatever
            unsigned integer this represents (assuming network-order),
            take that many bytes from the recovery bit string and
            append them to the new packet. This represents the CSRC
            list, extension, payload, and padding.

Rosenberg & Schulzrinne     Standards Track                    [Page 11]

RFC 2733                      Generic FEC                  December 1999

       15.  Set the SSRC of the new packet to the SSRC of the media
            stream it's protecting.

   This procedure will completely recover both the header and payload of
   an RTP packet.

8.2 Determination of When to Recover

   The previous section discussed how to recover a media packet with
   sequence number xi when all of the packets needed to recover it were
   available. The decision about whether to attempt recovery of some
   media packet xi, and how to determine if sufficient data is available
   to recover it, is left to the implementer. However, this section
   provides a simple algorithm which MAY be used for this purpose.

   The algorithm is described below in C code. The code assumes that
   several functions exist. recover_packet() takes the sequence number
   of a packet, and an FEC packet. Using the FEC packet and data packets
   received previously, the data packet with the given sequence number
   is recovered. add_fec_to_pending_list() adds the given FEC packet to
   a linked list of FEC packets which have not yet been used for
   recovery. wait_for_packet() waits for a packet, FEC or data, from the
   network. remove_from_pending_list() removes the FEC packet from the
   pending list. The structure packet contains a boolean variable fec
   which is true when the packet is FEC, false if it's media. When its
   an FEC packet, the mask and snbase field contain those values from
   the FEC packet header. When it's a media packet, the sn variable
   contains the sequence number of the packet. The global array A
   indicates which media packets have been received, and which have not.
   It is indexed by the sequence number of the packet.

   The function fec_recovery implements the algorithm. It waits for
   packets, and when it receives an FEC packet, calls recover_with_fec()
   to attempt to use it to recover. If no recovery is possible, the FEC
   packet is stored for later attempts. If the received packet was a
   media packet, its presence is noted, and any old FEC packets are
   checked to see if recovery is now possible. Recovered packets are
   treated as if they were received, triggering further attempts at

   A real implementation will need to use a circular buffer instead of
   the simple array (A in the code) in order to avoid running off the
   end of the buffer. In addition, the code below does not attempt to
   free up FEC packets that are old and were never used. Normally, such
   discarding is done based on time constraints introduced by the
   playout buffer. If an FEC data protects packets whose play time has
   elapsed, the FEC is no longer needed.

Rosenberg & Schulzrinne     Standards Track                    [Page 12]

RFC 2733                      Generic FEC                  December 1999

typedef struct packet_s {

  BOOLEAN fec;               /* FEC or media */

  int sn;                    /* SN of the packet, for media only */

  BOOLEAN mask[24];          /* Mask, FEC only */
  int snbase;                /* SN Base, FEC only */

  struct packet_s *next;

} packet;

BOOLEAN A[65535];
packet *pending_list;

packet *recover_with_fec(packet *fec_pkt) {

  packet *data_pkt;
  int pkts_present,  /* number of packets from the mask that are
                        present */
    pkts_needed,    /* number of packets needed is the number of ones
                        in the mask minus 1 */
    pkt_to_recover, /* sn of the packet we are recovering */

  pkts_present = 0;

  /* The number of packets needed is the number of ones in the mask
     minus 1.  The code below increments pkts_needed by the number
     of ones in the mask, so we initialize this to -1 so that the
     final count is correct */

  pkts_needed = -1;

  /* Go through all 24 bits in the mask, and check if we have
     all but one of the media packets */

  for(i = 0; i < 24; i++) {

     /* If the packet is here and in the mask, increment counter */

     if(A[i+fec_pkt->snbase] && fec_pkt->mask[i]) pkts_present++;

     /* Count the number of packets needed as well */
     if(fec_pkt->mask[i]) pkts_needed++;

Rosenberg & Schulzrinne     Standards Track                    [Page 13]

RFC 2733                      Generic FEC                  December 1999

     /* The packet to recover is the one with a bit in the
        mask that's not here yet */
     if(!A[i+fec_pkt->snbase] && fec_pkt->mask[i])
       pkt_to_recover = i+fec_pkt->snbase;

   /* If we can recover, do so. Otherwise, return NULL */

   if(pkts_present == pkts_needed) {
     data_pkt = recover_packet(pkt_to_recover, fec_pkt);
   }  else {
     data_pkt = NULL;


 void fec_recovery() {

   packet *p,    /* packet received or regenerated */
       *fecp,    /* fec packet from pending list */
       *pnew;    /* new packets recovered */

   while(1) {

     p = wait_for_packet();    /* get packet from network */

     while(p) {

       /* if it's an FEC packet, try to recover with it. If we can't,
          store it for later potential use. If we can recover, act as
          if the recovered packet is received and try to recover some
          more.  Otherwise, if it's a data packet, mark it as received,
          and check if we can now recover a data packet with the list
          of pending FEC packets */

       if(p->fec == TRUE) {
          pnew = recover_with_fec(p);


            A[pnew->sn] = TRUE;

          /* We assign pnew to p since the while loop will continue
             to recover based on p not being NULL */

Rosenberg & Schulzrinne     Standards Track                    [Page 14]

RFC 2733                      Generic FEC                  December 1999

          p = pnew;

       } else {

         /* Mark this data packet as here */
         A[p->sn] = TRUE;

         p = NULL;

         /* Go through pending list. Try and recover a packet using
            each FEC. If we are successful, add the data packet to
            the list of received packets, remove the FEC packet from
            the pending list, since we've used it, and then try to
            recover some more */

         for(fecp = pending_list; fecp != NULL; fecp = fecp->next) {
           pnew = recover_with_fec(fecp);
           if(pnew) {

             /* The packet is now here, as we've recovered it */
             A[pnew->sn] = TRUE;

             /* One FEC packet can only be used once to recover,
                so remove it from the pending list */


             p = pnew;


         } /*for*/

       } /*p->fec was false */

     } /* while p*/

   } /* while 1 */


Rosenberg & Schulzrinne     Standards Track                    [Page 15]

RFC 2733                      Generic FEC                  December 1999

9 Example

   Consider 2 media packets to be sent, x and y, from SSRC 2. Their
   sequence numbers are 8 and 9, respectively, with timestamps of 3 and
   5, respectively. Packet x uses payload type 11, and packet y uses
   payload type 18. Packet x is has 10 bytes of payload, and packet y
   11. Packet y has its marker bit set. The RTP headers for packets x
   and y are shown in Figures 3 and 4 respectively.

Media Packet x

   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
   |1 0|0|0|0 0 0 0|0|0 0 0 1 0 1 1|0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0|
   |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1|
   |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0|

      Version:   2
      Padding:   0
      Extension: 0
      Marker:    0
      PTI:       11
      SN:        8
      TS:        3
      SSRC:      2

   Figure 3: RTP Header for Media Packet X

   An FEC packet is generated from these two. We assume that payload
   type 127 is used to indicate an FEC packet. The resulting RTP header
   is shown in Figure 5.

   The FEC header in the FEC packet is shown in Figure 6.

Rosenberg & Schulzrinne     Standards Track                    [Page 16]

RFC 2733                      Generic FEC                  December 1999

11 Use with Redundant Encodings

   One can consider an FEC packet as a "redundant coding" of the media.

Media Packet y

   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
   |1 0|0|0|0 0 0 0|1|0 0 1 0 0 1 0|0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1|
   |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1|
   |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0|

      Version:   2
      Padding:   0
      Extension: 0
      Marker:    1
      PTI:       18
      SN:        9
      TS:        5
      SSRC:      2

   Figure 4: RTP Header for Media Packet Y

   Because of this, the payload format for encoding of redundant audio
   data [5] can be used to carry the FEC data along with the media. The
   procedure for this is as follows.

   The FEC operation defined above acts on a stream of RTP media
   packets. The stream which is operated on is the stream before the
   encapsulation defined in RFC 2198 [5]. In other words, the media
   stream to be protected is encapsulated in standard RTP media packets.
   The FEC operation above is performed (with one minor change),
   generating a stream of FEC packets. The change to the procedure above
   is that if the RTP packets being protected contain an RTP extension,
   padding, or a CSRC list, these MUST be removed from the packets, and
   the CC field, Padding Bit, and Extension but MUST be set to zero,
   before the FEC operation is applied. These modified packets are used
   in the procedure above. Note that the sender MUST still send the
   original packets (with the CSRC list, padding, and extension in tact)
   as the primary encoding in RFC 2198. The removal of these fields only
   applies to the protection operation.

Rosenberg & Schulzrinne     Standards Track                    [Page 17]

RFC 2733                      Generic FEC                  December 1999

   Once the FEC packets have been generated, the media payload is
   extracted from the media packets. This payload is used as the primary
   encoding as defined in RFC 2198. Then, the FEC header and payload of
   the FEC packets is extracted, and treated as a redundant encoding.
   Additional redundant encodings, besides FEC, MAY be added to the
   packet as well. These encodings will not be protected by FEC,

   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
   |1 0|0|0|0 0 0 0|1|1 1 1 1 1 1 1|0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1|
   |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1|
   |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0|

      Version:   2
      Padding:   0
      Extension: 0
      Marker:    1
      PTI:       127
      SN:        1
      TS:        5
      SSRC:      2

   Figure 5: RTP Header of FEC for Packets X and Y

   The redundant encodings header for the primary codec is set as
   defined in RFC 2198. The redundant encodings header for the FEC data
   is set as follows. The block PT is set to the dynamic PT associated
   with the FEC format. The block length is set to the sum of the
   lengths of the FEC header and payload. The timestamp offset SHOULD be
   set to zero. The secondary coder payload includes the FEC header and
   FEC payload.

   At the receiver, the primary codec and all secondary codecs are
   extracted as separate RTP packets. This is done by copying the
   sequence number, SSRC, marker bit, CC field, RTP version, and
   extension bit from the RTP header of the redundant encodings packet
   to the RTP header of each extracted packet. If the secondary codec
   contains FEC, the CC field, Extension Bit, and Padding Bit in the RTP
   header of the FEC packet MUST be set to zero instead. The payload
   type identifier in the extracted packet is copied from the block PT
   of the redundant encodings header. The timestamp of the extracted
   packet is the difference between the timestamp in the RTP header and

Rosenberg & Schulzrinne     Standards Track                    [Page 18]

RFC 2733                      Generic FEC                  December 1999

   the offset in the block header. The payload of the extracted packet
   is the data block. This will result in the FEC stream and media
   stream being extracted.

   |0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0|0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1|
   |0|0 0 1 1 0 0 1|0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1|
   |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0|

      SN base:   8    [min(8,9)]
      len. rec.: 1    [8 xor 9]
      E:         0
      PTI rec.:  25   [11 xor 18]
      mask:      3
      TS rec.:   6    [3 xor 5]

      The payload length is 11 bytes.

   Figure 6: FEC Header of Result

   To use the FEC and media packets for recovery, the CSRC list,
   extension, and padding MUST be removed from the media packets, if
   present, and the CC field, Extension Bit, and Padding Bit MUST be set
   to zero. These modified media packets, along with the FEC packets,
   are then used to recover based on the procedures in section 8. The
   recovered media packets will always have no extension, padding, or
   CSRC list. An implementation MAY copy these fields into the recovered
   packet from another media packet, if available.

   Using the redundant encodings payload format also implies that the
   marker bit may not be recovered correctly. Applications MUST set the
   marker bit to zero in media packets reconstructed using FEC
   encapsulated in RFC 2198 redundancy.

   An advantage of this approach is a reduction in the overhead for
   sending FEC packets.

Rosenberg & Schulzrinne     Standards Track                    [Page 19]

RFC 2733                      Generic FEC                  December 1999

11 Indicating FEC Usage in SDP

   FEC packets contain RTP packets with dynamic payload type values. In
   addition, the FEC packets can be sent on separate multicast groups or
   separate ports from the media. The FEC can even be carried in packets
   containing media, using the redundant encodings payload format [5].
   These configuration options must be indicated out of band. This
   section describes how this can be accomplished using the Session
   Description Protocol (SDP), specified in RFC 2327 [6].

11.1 FEC as a Separate Stream

   In the first case, the FEC packets are sent as a separate stream.
   This can mean they are sent on a different port and/or multicast
   group from the media. When this is done, several pieces of
   information must be conveyed:

        o The address and port where the FEC is being sent to

        o The payload type number for the FEC

        o Which media stream the FEC is protecting

   The payload type number for the FEC is conveyed in the m line of the
   media it is protecting, listed as if it were another valid encoding
   for the stream. There is no static payload type assignment for FEC,
   so dynamic payload type numbers MUST be used. The binding to the
   number is indicated by an rtpmap attribute. The name used in this
   binding is "parityfec".

   The presence of the payload type number in the m line of the media it
   is protecting does not mean the FEC is sent to the same address and
   port as the media. Instead, this information is conveyed through an
   fmtp attribute line. The presence of the FEC payload type on the m
   line of the media serves only to indicate which stream the FEC is

   The format for the fmtp line for FEC is:

   a=fmtp:<number> <port> <network type> <addresss type> <connection

   where 'number' is the payload type number present in the m line. Port
   is the port number where the FEC is sent to. The remaining three
   items - network type, address type, and connection address - have the
   same syntax and semantics as the c line from SDP. This allows the
   fmtp line to be partially parsed by the same parser used on the c

Rosenberg & Schulzrinne     Standards Track                    [Page 20]

RFC 2733                      Generic FEC                  December 1999

   lines. Note that since FEC cannot be hierarchically encoded, the
   <number of addresses> parameter MUST NOT appear in the connection

   The following is an example SDP for FEC:

   o=hamming 2890844526 2890842807 IN IP4
   s=FEC Seminar
   c=IN IP4
   t=0 0
   m=audio 49170 RTP/AVP 0 78
   a=rtpmap:78 parityfec/8000
   a=fmtp:78 49172 IN IP4
   m=video 51372 RTP/AVP 31 79
   a=rtpmap:79 parityfec/8000
   a=fmtp:79 51372 IN IP4

   The presence of two m lines in this SDP indicates that there are two
   media streams - one audio and one video. The media format of 0
   indicates that the audio uses PCM, and is protected by FEC with
   payload type number 78. The FEC is sent to the same multicast group
   and TTL as the audio, but on a port number two higher (49172). The
   video is protected by FEC with payload type number 79. The FEC
   appears on the same port as the video (51372), but on a different
   multicast address.

11.2 Use with Redundant Encodings

   When the FEC stream is being sent as a secondary codec in the
   redundant encodings format, this must be signaled through SDP. To do
   this, the procedures defined in RFC 2198 are used to signal the use
   of redundant encodings. The FEC payload type is indicated in the same
   fashion as any other secondary codec. An rtpmap attribute MUST be
   used to indicate a dynamic payload type number for the FEC packets.
   The FEC MUST protect only the main codec. In this case, the fmtp
   attribute for the FEC MUST NOT be present.

   For example:

   m=audio 12345 RTP/AVP 121 0 5 100
   a=rtpmap:121 red/8000/1
   a=rtpmap:100 parityfec/8000
   a=fmtp:121 0/5/100

Rosenberg & Schulzrinne     Standards Track                    [Page 21]

RFC 2733                      Generic FEC                  December 1999

   This SDP indicates that there is a single audio stream, which can
   consist of PCM (media format 0) , DVI (media format 5), the redundant
   encodings (indicated by media format 121, which is bound to red
   through the rtpmap attribute), or FEC (media format 100, which is
   bound to parityfec through the rtpmap attribute). Although the FEC
   format is specified as a possible coding for this stream, the FEC
   MUST NOT be sent by itself for this stream. Its presence in the m
   line is required only because non-primary codecs must be listed here
   according to RFC 2198. The fmtp attribute indicates that the
   redundant encodings format can be used, with DVI as a secondary
   coding and FEC as a tertiary encoding.

11.3 Usage with RTSP

   RTSP [7] can be used to request FEC packets to be sent as a separate
   stream. When SDP is used with RTSP, the Session Description does not
   include a connection address and port number for each stream.
   Instead, RTSP uses the concept of a "Control URL". Control URLs are
   used in SDP in two distinct ways.

        1.   There is a single control URL for all streams. This is
             referred to as "aggregate control". In this case, the fmtp
             line for the FEC stream is omitted.

        2.   There is a Control URL assigned to each stream. This is
             referred to as "non-aggregate control". In this case, the
             fmtp line specifies the Control URL for the stream of FEC
             packets. The URL may be used in a SETUP command by an RTSP

   The format for the fmtp line for FEC with RTSP and non-aggregate
   control is:

   a=fmtp:<number> <control URL>

   where 'number' is the payload type number present in the m line.
   Control URL is the URL used to control the stream of FEC packets.
   Note that the Control URL does not need to be an absolute URL. The
   rules for converting a relative Control URL to an absolute URL are
   given in RFC 2326, Section C.1.1.

Rosenberg & Schulzrinne     Standards Track                    [Page 22]

RFC 2733                      Generic FEC                  December 1999

12 Security Considerations

   The use of FEC has implications on the usage and changing of keys for
   encryption. As the FEC packets do consist of a separate stream, there
   are a number of permutations on the usage of encryption. In

     o The FEC stream may be encrypted, while the media stream is

     o The media stream may be encrypted, while the FEC stream is

     o The media stream and FEC stream are both encrypted, but using
       different keys.

     o The media stream and FEC stream are both encrypted, but using
       the same key.

   The first three of these would require any application level
   signaling protocols to be aware of the usage of FEC, and to thus
   exchange keys for it and negotiate its usage on the media and FEC
   streams separately. In the final case, no such additional mechanisms
   are needed. The first two cases present a layering violation, as FEC
   packets should really be treated no differently than other RTP
   packets. Encrypting just one may also make certain known-plaintext
   attacks possible. For these reasons, applications utilizing
   encryption SHOULD encrypt both streams.

   However, the changing of keys becomes problematic. For example, if
   two packets a and b are sent, and FEC packet f(a,b) is sent, and the
   keys used for a and b are different, which key should be used to
   decode f(a,b)? In general, old keys will likely need to be cached, so
   that when the keys change for the media stream, the old key is kept,
   and used, until it is determined that the key has changed on the FEC
   packets as well.

   Another issue with the use of FEC is its impact on network
   congestion. Adding FEC in the face of increasing network losses is a
   bad idea, as it can lead to increased congestion and eventual
   congestion collapse if done on a widespread basis. As a result,
   implementers MUST NOT substantially increase the amount of FEC in use
   as network losses increase.

Rosenberg & Schulzrinne     Standards Track                    [Page 23]

RFC 2733                      Generic FEC                  December 1999

13 Acknowledgments

   This work is based on an earlier draft on FEC, submitted by Budge and
   Mackenzie in 1997. We would also like to thank Steve Casner, Mark
   Handley, Orion Hodson and Colin Perkins for their comments. Thanks to
   Anders Klemets who wrote the section on usage with RTSP.

14 Authors' Addresses

   Jonathan Rosenberg
   200 Executive Drive
   Suite 120
   West Orange, NJ 07046

   Email: jdrosen@dynamicsoft.com

   Henning Schulzrinne
   Columbia University
   M/S 0401, 1214 Amsterdam Ave.
   New York, NY 10027-7003

   EMail: schulzrinne@cs.columbia.edu

Rosenberg & Schulzrinne     Standards Track                    [Page 24]

RFC 2733                      Generic FEC                  December 1999

15 Bibliography

   [1] J.C. Bolot and A. V. Garcia, "Control mechanisms for packet audio
       in the internet," in Proceedings of the Conference on Computer
       Communications (IEEE Infocom) , (San Francisco, California), Mar.

   [2] Perkins, C. and O. Hodson, "Options for Repair of Streaming
       media", RFC 2354, June 1998.

   [3] Schulzrinne, H., Casner, S., Frederick, R. and V. Jacobson, "RTP:
       A Transport Protocol for Real-Time Applications", RFC 1889,
       January 1996.

   [4] Bradner, S., "Key words for use in RFCs to indicate requirement
       levels", BCP 14, RFC 2119, March 1997.

   [5] Perkins, C., Kouvelas, I., Hodson, O., Hardman, V., Handley, M.,
       Bolot, J.C., Vega-Garcia, A. and S. Fosse-Parisis, "RTP Payload
       for Redundant Audio Data", RFC 2198, September 1997.

   [6] Handley, M. and V. Jacobson, "SDP: Session Description Protocol",
       RFC 2327, April 1998.

   [7] Schulzrinne, H., Rao, A. and R. Lanphier, "Real Time Streaming
       Protocol (RTSP)", RFC 2326, April 1998.

Rosenberg & Schulzrinne     Standards Track                    [Page 25]

RFC 2733                      Generic FEC                  December 1999

16.  Full Copyright Statement

   Copyright (C) The Internet Society (1999).  All Rights Reserved.

   This document and translations of it may be copied and furnished to
   others, and derivative works that comment on or otherwise explain it
   or assist in its implementation may be prepared, copied, published
   and distributed, in whole or in part, without restriction of any
   kind, provided that the above copyright notice and this paragraph are
   included on all such copies and derivative works.  However, this
   document itself may not be modified in any way, such as by removing
   the copyright notice or references to the Internet Society or other
   Internet organizations, except as needed for the purpose of
   developing Internet standards in which case the procedures for
   copyrights defined in the Internet Standards process must be
   followed, or as required to translate it into languages other than

   The limited permissions granted above are perpetual and will not be
   revoked by the Internet Society or its successors or assigns.

   This document and the information contained herein is provided on an


   Funding for the RFC Editor function is currently provided by the
   Internet Society.

Rosenberg & Schulzrinne     Standards Track                    [Page 26]