Internet Engineering Task Force                                   AVT WG
INTERNET-DRAFT                                              Ladan Gharai
draft-ietf-avt-tfrc-profile-07.txt                               USC/ISI
                                                            1 March 2007
                                                 Expires: September 2007

                   RTP with TCP Friendly Rate Control

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

   The list of Internet-Draft Shadow Directories can be accessed at

Copyright Notice

   Copyright (C) The Internet Society (2007).


   This memo specifies how the TCP Friendly Rate Control (TFRC) of RTP
   flows can be supported using the RTP/AVPF profile and the general RTP
   header extension mechanism.  AVPF feedback packets and RTP header
   extensions are defined to support the exchange of control information
   between RTP TFRC senders and receivers. TFRC is an equation based
   congestion control scheme for unicast flows operating in a best
   effort Internet environment.

Gharai                                                          [Page 1]

INTERNET-DRAFT           Expires: September 2007              March 2007

1.  Introduction

   [Note to RFC Editor: All references to RFC XXXX are to be replaced
   with the RFC number of this memo, when published]

   This memo specifies how the TCP Friendly Rate Control (TFRC) of RTP
   flows can be supported using the RTP/AVPF profile and RTP header
   extensions, by defining the header extensions to be used and a new
   AVPF feedback packet.

   TFRC is an equation based congestion control scheme for unicast flows
   operating in a best effort Internet environment and competing with
   TCP traffic. TFRC computes a TCP-friendly data rate based on current
   network conditions, as represented by the latest round trip time and
   packet loss calculations. The complete TFRC mechanism is described in
   detail in [TFRC].

   To calculate a TCP-friendly data rate and keep track of round trip
   times and packet losses, TFRC senders and receivers rely on
   exchanging specific information between each other, i.e: the sender
   provides the receiver with the latest updates to round trip time
   calculations, while the receiver provides feedback needed to compute
   round trip times and on packet losses. This memo defines how this
   information can be exchanged between TFRC senders and receiver with
   RTP header extensions and an AVPF feedback packet.

2.  Terminology

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

3.  Relation to the Datagram Congestion Control Protocol

   The TFRC congestion control mechanism is also supported by the
   Datagram Congestion Control Protocol (DCCP). In this section we
   detail the pros and cons of using TFRC with RTP versus DCCP.

   DCCP is a minimal general purpose transport-layer protocol with
   unreliable yet congestion controlled packet delivery semantics and
   reliable connection setup and teardown. DCCP currently supports both
   TFRC and TCP-like congestion control, and the protocol is structured
   to support new congestion control mechanisms defined in the future.
   In addition DCCP supports a host of other features, such as: use of
   Explicit Congestion Notification (ECN) and the ECN Nonce, reliable
   option negotiation and Path Maximum Transfer Unit (PMTU).  Naturally
   an application using RTP/DCCP as its transport protocol will benefit
   from the protocol features supported by DCCP.

Gharai                                                          [Page 2]

INTERNET-DRAFT           Expires: September 2007              March 2007

   However there are a number of benefits to be gained by the
   development and standardization of the use RTP with TFRC:

     o Media applications lacking congestion control can incorporate
       congestion controlled transport without delay by using RTP with
       TFRC. The DCCP protocol is currently under development and
       widespread deployment is not yet in place.

     o Use of the RTP with TFRC is not contingent on any OS level
       changes and can be quickly deployed, as RTP is implemented
       at the application layer.

     o RTP/UDP flows face the same restrictions in firewall traversal
       as do UDP flows and do not require NATs and firewall
       modifications.   DCCP flows, on the other hand, do require NAT
       and firewall modifications, however once these modifications are
       in place, they can result in easier NAT and firewall traversal
       for RTP/DCCP flows in the future.

     o Use of RTP with TFRC with various media applications will give
       researchers, implementors and developers a better understanding
       of the intricate relationship between media quality and equation
       based congestion control.  Hopefully this experience with
       congestion control and TFRC will ease the migration of media
       applications to DCCP once DCCP is deployed.

   Overall, using the AVPF/RTP profile and header extension to support
   TFRC provides an immediate means for congestion control in media
   streams, in the time being until DCCP is deployed.

   Additionally, there are also a number of technical differences as to
   how (and which) congestion control information is exchanged between
   DCCP with CCID3 and RTP:

     o Using header extensions the RTP TFRC sender transmits a
       send timestamp to the RTP TFRC receiver with every data packet.
       In addition to congestion control the send timestamp can be
       used by the receiver for jitter calculations.

       In contrast DCCP with CCID3 transmits a quad round trip
       counter to the receiver.

     o The RTP TFRC receiver only provides the RTP TFRC sender
       with the loss event rate as computed by the receiver.

       In contrast DCCP with CCID3, provides 2 other options for the
       transport of loss event rate. A sender may choose to receive
       loss intervals or an Ack Vector. These two options provide the

Gharai                                                          [Page 3]

INTERNET-DRAFT           Expires: September 2007              March 2007

       sender with the necessary information to compute the loss event

     o Sequence number: DCCP supports a 48 bit and a 24 bit sequence
       number, whereas RTP only supports a 16 bit sequence number. While
       this makes RTP susceptible to data injection attacks, it can be
       avoided by using the SRTP [SRTP] profile.

4.  The TFRC Information Exchange Loop

   TFRC depends on the exchange of congestion control information
   between a sender and receiver.  In this section we reiterate which
   items are exchanged between a TFRC sender and receiver as discussed
   in [TFRC]. We note how the RTP can accommodates these exchanges.

4.1.  Data Packets

   As stated in [TFRC] a TFRC sender transmits the following information
   in each data packet to the receiver:

    o A sequence number, incremented by one for each data packet

    o A timestamp indicating the packet send time and the sender's
      current estimate of the round-trip time, RTT. This information
      is then used by the receiver to compute the TFRC loss intervals.
      - or -
      A course-grained timestamp incrementing every quarter of a
      round trip time, which is then used to determine the TFRC loss

   The standard RTP sequence number suffices for TFRCs functionality.
   RTP header extension [hdrtxt] are used to transmit the send timestamp
   and RTT.   The RTT can be transmitted in band with every RTP packet
   or when there is significant change is the RTT. Each extension
   payload is 3 bytes long (see Section 6).

4.2.  Feedback Packets

   As stated in [TFRC] a TFRC receiver provides the following feedback
   to the sender at least once per RTT or per data packet received
   (which ever time interval is larger):

    o The send timestamp of the last data packet received, t_i.

    o The amount of time elapsed between the receipt of the last
      data packet at the receiver, and the generation of this feedback

Gharai                                                          [Page 4]

INTERNET-DRAFT           Expires: September 2007              March 2007

      report, t_delay. This is used by the sender for RTT computations.

    o The rate at which the receiver estimates that data was received
      since the last feedback report was sent, x_recv.

    o The receiver's current estimate of the loss event rate, p, a real
      value between 0 and 1.0.

   To accommodate the feedback of these values a new AVPF transport
   layer feedback message is defined, as detailed in Section 7.

5.  The Header Extensions

   The form of the extension block when both the RTT and send timestamp
   are being transmitted is depicted in Figure 1.  The length field for
   each extensions takes the value 2 to indicate that the payload is 3
   bytes. The two header extension fields are defined and used as

   Send timestamp: 24 bits
     The timestamp indicating when the packet is sent. This timestamp
     is measured in microseconds and is used for round trip time

   Round trip time (RTT): 24 bits
     The round trip time as measured by the RTP TFRC sender in

       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
      |      0xBE     |      0xDE     |            length=2           |
      |  ID   | len=2 |                RTT                            |
      |  ID   | len=2 |          send timestamps                      |
     Figure 1

6.  TFRC-FB: A New AVPF Transport Layer Feedback Message

   To support feedback to the  receivers a new transport layer AVPF
   feedback message is defined: TFRC-FB. This message is depicted in
   Figure 2.  It is defined according to [AVPF] and includes the
   following four fields:

Gharai                                                          [Page 5]

INTERNET-DRAFT           Expires: September 2007              March 2007

   Timestamp (t_i): 32 bits
     The send timestamp of the last data packet received by the
     RTP TFRC receiver, t_i, in microseconds.

   Delay (t_delay): 32 bits
     The amount of time elapsed between the receipt of the last data
     packet at the RTP TFRC receiver, and the generation of this
     feedback report in microseconds. This is used by the RTP TFRC
     sender for RTT computations.

   Data rate (x_recv): 32 bits
     The rate at which the receiver estimates that data was received
     since the last feedback report was sent in bytes per second.

   Loss event rate (p): 32 bits
     The receiver's current estimate of the loss event rate, p,
     expressed as a fixed point number with the binary point at the
     left edge of the field. (That is equivalent to taking the integer
     part after multiplying the loss event rate by 2^32.)

      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
      |V=2|P|  FMT=2  |  PT=RTPFB     |             length            |
      |                     SSRC of packet sender                     |
      |                   SSRC (SSRC of first source)                 |
      |                             t_i                               |
      |                           t_delay                             |
      |                  data rate at the receiver (x_recv)           |
      |                    loss event rate (p)                        |
     Figure 2

7.  RTCP Transmission Intervals, TFRC and the AVPF Profile

   When running TFRC rate controlled RTP, the RTCP transmission
   intervals MUST be set according to the requirements of the TFRC
   algorithm. TFRC requires a receiver to generate a feedback packet at
   least once per RTT or per packet received (based on the larger time

Gharai                                                          [Page 6]

INTERNET-DRAFT           Expires: September 2007              March 2007

   interval). These requirements are to ensure timely reaction to

   To support the transmission of feedback packet once per RTT, a
   RTP/AVPF flow with TFRC congestion control must:

    o set allow_early to "true" at all times. Essentially, this means
      that a receiver can always generate an early feedback packet, and
      does not need to alternate between early and regular RTCP packets
      (see RFC 4585, Section 3.4,k).

    o T_rr_interval must not be set to a value larger than the current
      round trip time, as this would prevent generating feedback packets
      at least once per RTT (see RFC 4585, Section 3.4,m).

   The TFRC requirements of receiving feedback once per RTT can at times
   conflict with the AVP RTCP bandwidth constraints, particularly at
   small RTTs of 20ms or less.  Assuming only one TFRC-FB report per
   RTCP compound packet, Table 1 lists the RTCP bandwidths at RTTs of 2,
   5, 10 and 20 ms and the minimum corresponding RTP data rates, where
   RTCP(X) <= (0.05)*RTP(X) is true.   For example, according to Table
   1, a TFRC RTP flow of less than 3.2 Mbps and a RTT of 5 ms, can not
   comply with the 5% RTCP bandwidth constraints (Table 1 assumes each
   RTCP packet is 100 bytes). RTP flows facing such circumstance should
   take into account the additional RTCP bandwidth needed when signaling
   their bandwidth information in SDP.

                        RTT      RTCP(X)   RTP(X)
                    |  20 ms |  40 kbps | 0.8 Mbps |
                    |  10 ms |  80 kbps | 1.6 Mbps |
                    |   5 ms | 160 kbps | 3.2 Mbps |
                    |   2 ms | 400 kbps | 8.0 Mbps |
                                Table 1

8.  SDP Definitions

   RTP flows using TFRC congestion control must signal their use of
   header extensions for round trip times (RTT) and the send timestamp:

     a=extmap:4 urn:ietf:params:rtp-hdtext:rtt
     a=extmap:4 urn:ietf:params:rtp-hdtext:send-ts

Gharai                                                          [Page 7]

INTERNET-DRAFT           Expires: September 2007              March 2007

9.  IANA Considerations

   In this section we detail IANA registry values that need to be

   The new RTP/AVPF feedback packet, TFRC-FB, must be registered. For
   the RTPFB range of packets, the following format (FMT) values are

     Value name:  TFRC-FB
     Long name:   TFRC feedback
     Value:  2
     Reference:   RFC XXXX

   The names rtt and send-ts need to be registered into the rtp-hdrext
   section of the urn:ietf: namespace, referring to RFC XXXX.

10.  Security Considerations

   This memo defines how to use the RTP AVPF profile and the general RTP
   header extensions to support TFRC congestion control. Therefore RTP
   packets using these mechanisms are subject to the security
   considerations discussed in the RTP specification [RTP], the RTP/AVPF
   profile specification [AVPF] and the general header extensions
   mechanism [hdrtxt]. Combining these mechanisms does not pose any
   additional security implications.  Applications requiring
   authentication and integrity protection can use the SAVPF [SAVPF]

11.  Acknowledgments

   This memo is based upon work supported by the U.S. National Science
   Foundation (NSF) under Grant No. 0334182. Any opinions, findings and
   conclusions or recommendations expressed in this material are those
   of the authors and do not necessarily reflect the views of NSF.

12.  Author's Address

     Ladan Gharai <>
     USC Information Sciences Institute
     3811 N. Fairfax Drive, #200
     Arlington, VA 22203

Gharai                                                          [Page 8]

INTERNET-DRAFT           Expires: September 2007              March 2007

Normative References

   [RTP]    H. Schulzrinne, S. Casner, R. Frederick and V. Jacobson,
            "RTP: A Transport Protocol for Real-Time Applications",
            Internet Engineering Task Force, RFC 3550 (STD0064), July

   [AVP]    H. Schulzrinne and S. Casner, "RTP Profile for Audio and
            Video Conferences with Minimal Control," RFC 3551 (STD0065),
            July 2003.

   [AVPF]   J. Ott, S. Wenger, A. Sato, C. Burmeister and J. Ray,
            "Extended RTP Profile for RTCP-based Feedback (RTP/AVPF)",
         RFC 4585, July 2006.

   [2119]   S. Bradner, "Key words for use in RFCs to Indicate
            Requirement Levels", Internet Engineering Task Force,
            RFC 2119, March 1997.

   [2434]   T. Narten and H. Alvestrand, "Guidelines for Writing an IANA
            Considerations Section in RFCs", Internet Engineering Task
            Force, RFC 2434, October 1998.

   [TFRC]   M. Handley, S. Floyed, J. Padhye and J. widmer,
            "TCP Friendly Rate Control (TRFC): Protocol Specification",
            Internet Engineering Task Force, RFC 3448, January 2003.

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

   [SRTP]   M. Baugher, D. McGrew, M. Naslund, E. Carrara, K.  Norrman,
            "The Secure Real-time Transport Protocol", RFC 3711, March

   [hdrext] D. Singer, "A general mechanism for RTP Header Extensions",
         ID draft-ietf-avt-rtp-hdrext-08, October 2006.

   [SAVPF]  J. Ott, E. Carrara, "Extended Secure RTP Profile for
            RTCP-based Feedback (RTP/SAVPF)," draft-ietf-avt-profile-
         savpf-09.txt, April, 2007.

Informative References

13.  IPR Notice

   The IETF takes no position regarding the validity or scope of any
   Intellectual Property Rights or other rights that might be claimed to

Gharai                                                          [Page 9]

INTERNET-DRAFT           Expires: September 2007              March 2007

   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

   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-

14.  Full Copyright Statement

   Copyright (C) The IETF Trust (2007).

   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.

   This document and the information contained herein are provided on an

Gharai                                                         [Page 10]