RFC 6284






Internet Engineering Task Force (IETF)                          A. Begen
Request for Comments: 6284                                       D. Wing
Category: Standards Track                                          Cisco
ISSN: 2070-1721                                          T. Van Caenegem
                                                          Alcatel-Lucent
                                                               June 2011


        Port Mapping between Unicast and Multicast RTP Sessions

Abstract



   This document presents a port mapping solution that allows RTP
   receivers to choose their own ports for an auxiliary unicast session
   in RTP applications using both unicast and multicast services.  The
   solution provides protection against denial-of-service or packet
   amplification attacks that could be used to cause one or more RTP
   packets to be sent to a victim client.

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 5741.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at
   http://www.rfc-editor.org/info/rfc6284.

Copyright Notice



   Copyright (c) 2011 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
   (http://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.




Begen, et al.                Standards Track                    [Page 1]

RFC 6284                      Port Mapping                     June 2011


Table of Contents



   1.  Introduction . . . . . . . . . . . . . . . . . . . . . . . . .  3
   2.  Requirements Notation  . . . . . . . . . . . . . . . . . . . .  4
   3.  Token-Based Port Mapping . . . . . . . . . . . . . . . . . . .  5
     3.1.  Motivating Scenario  . . . . . . . . . . . . . . . . . . .  6
     3.2.  Normative Behavior and Requirements  . . . . . . . . . . .  9
   4.  Message Formats  . . . . . . . . . . . . . . . . . . . . . . . 11
     4.1.  Port Mapping Request . . . . . . . . . . . . . . . . . . . 12
     4.2.  Port Mapping Response  . . . . . . . . . . . . . . . . . . 13
     4.3.  Token Verification Request . . . . . . . . . . . . . . . . 15
       4.3.1.  Where to Include Token . . . . . . . . . . . . . . . . 16
     4.4.  Token Verification Failure . . . . . . . . . . . . . . . . 17
   5.  Procedures for Token Construction  . . . . . . . . . . . . . . 18
   6.  Validating Tokens  . . . . . . . . . . . . . . . . . . . . . . 20
   7.  SDP Signaling  . . . . . . . . . . . . . . . . . . . . . . . . 21
     7.1.  The 'portmapping-req' Attribute  . . . . . . . . . . . . . 21
       7.1.1.  ABNF Definition of 'portmapping-req' . . . . . . . . . 21
       7.1.2.  Offer/Answer Model Considerations  . . . . . . . . . . 22
     7.2.  Requirements . . . . . . . . . . . . . . . . . . . . . . . 22
     7.3.  Example and Discussion . . . . . . . . . . . . . . . . . . 23
   8.  Address Pooling NATs . . . . . . . . . . . . . . . . . . . . . 24
   9.  Security Considerations  . . . . . . . . . . . . . . . . . . . 24
     9.1.  Tokens . . . . . . . . . . . . . . . . . . . . . . . . . . 24
     9.2.  The 'portmapping-req' Attribute  . . . . . . . . . . . . . 26
   10. IANA Considerations  . . . . . . . . . . . . . . . . . . . . . 26
     10.1. Registration of SDP Attributes . . . . . . . . . . . . . . 26
     10.2. Registration of RTCP Control Packet Types  . . . . . . . . 27
     10.3. SMT Values for TOKEN Packet Type Registry  . . . . . . . . 27
     10.4. RAMS Response Code Space Registry  . . . . . . . . . . . . 27
   11. Acknowledgments  . . . . . . . . . . . . . . . . . . . . . . . 28
   12. References . . . . . . . . . . . . . . . . . . . . . . . . . . 28
     12.1. Normative References . . . . . . . . . . . . . . . . . . . 28
     12.2. Informative References . . . . . . . . . . . . . . . . . . 29

















Begen, et al.                Standards Track                    [Page 2]

RFC 6284                      Port Mapping                     June 2011


1.  Introduction



   In (any-source or source-specific) multicast RTP applications,
   destination ports (i.e., the ports on which the multicast receivers
   receive the RTP and RTP Control Protocol (RTCP) packets) are defined
   declaratively.  In other words, the receivers cannot choose their
   receive ports, and the sender(s) use the predefined ports.

   In unicast RTP applications, the receiving end needs to choose its
   ports for RTP and RTCP since these ports are local resources and only
   the receiving end can determine which ports are available to use.  In
   addition, Network Address Port Translation (NAPT, hereafter simply
   called NAT) devices are commonly deployed in networks; thus, static
   port assignments cannot be used.  The receiving end may convey its
   request to the sending end through different ways, one of which is
   the Offer/Answer Model [RFC3264] for the Session Description Protocol
   (SDP) [RFC4566].  However, the Offer/Answer Model requires offer/
   answer exchange(s) between the endpoints, and the resulting delay may
   not be desirable in delay-sensitive real-time applications.
   Furthermore, the Offer/Answer Model may be burdensome for the
   endpoints that are concurrently running a large number of unicast
   sessions with other endpoints.

   In this specification, we consider an RTP application that uses one
   or more unicast and multicast RTP sessions together.  While the
   declaration and selection of the ports are well defined and work well
   for multicast and unicast RTP applications, respectively, the usage
   of the ports introduces complications when a receiving end mixes
   unicast and multicast RTP sessions within the same RTP application.

   An example scenario is where the RTP packets are distributed through
   source-specific multicast (SSM) [RFC4607] and a receiver sends
   unicast RTCP NACK feedback [RFC4585] to a local repair server (also
   functioning as a unicast RTCP feedback target) [RFC5760] asking for a
   retransmission of the packets it is missing, and the local repair
   server sends the retransmission packets over a unicast RTP session
   [RETRANSMISSION-FOR-SSM].

   Another scenario is where a receiver wants to rapidly acquire a new
   primary multicast RTP session and receives one or more RTP burst
   packets over a unicast session before joining the SSM session; see
   [RFC6285] regarding Rapid Acquisition of Multicast RTP Sessions
   (RAMS).  Similar scenarios exist in applications where some part of
   the content is distributed through multicast while the receivers get
   additional and/or auxiliary content through one or more unicast
   connections, as illustrated in Figure 1.





Begen, et al.                Standards Track                    [Page 3]

RFC 6284                      Port Mapping                     June 2011


   In this document, we discuss this problem and introduce a solution
   that we refer to as port mapping.  This solution allows receivers to
   choose their desired UDP ports for RTP and RTCP in every unicast
   session when they are running RTP applications using both unicast and
   multicast services and offer/answer exchange is not available.  The
   solution includes a Token-based protection mechanism against denial-
   of-service (DoS) or packet amplification attacks that could be used
   to cause one or more RTP packets to be sent to a victim client.  This
   solution is not applicable in cases where TCP is used as the
   transport protocol in the unicast sessions.  For such scenarios,
   refer to [RFC4145].

          -----------
         |  Unicast  |................
         |  Source   |.............  :
         | (Server)  |            :  :
          -----------             :  :
                                  v  v
          -----------          ----------             -----------
         | Multicast |------->|  Router  |---------->|Client RTP |
         |  Source   |        |          |..........>|Application|
          -----------          ----------             -----------
                                   | :
                                   | :                -----------
                                   | :..............>|Client RTP |
                                   +---------------->|Application|
                                                      -----------


         -------> Multicast RTP Flow
         .......> Unicast RTP Flow

     Figure 1: RTP Applications Simultaneously Using Both Unicast and
                            Multicast Services

   In the remainder of this document, we refer to the RTP endpoints that
   serve other RTP endpoints over a unicast session as the Servers.  The
   receiving RTP endpoints are referred to as Clients.  This terminology
   also reflects the fact that when port mapping is used, the RTP
   packets can only flow in one direction (from the server to the
   client) in the unicast sessions.

2.  Requirements Notation



   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
   [RFC2119].



Begen, et al.                Standards Track                    [Page 4]

RFC 6284                      Port Mapping                     June 2011


3.  Token-Based Port Mapping



   Token-based port mapping consists of the server providing the client
   a Token that can be used to establish a unicast session without the
   possibility of an attacker redirecting traffic to an unsuspecting
   third party to create a DoS attack.  The Token is essentially an
   opaque encapsulation that is based on the client's IP address (as
   seen by the server), a time-to-live value, and a random nonce
   provided by the client.

   Token-based port mapping consists of two steps: (i) Token request and
   retrieval, and (ii) unicast session establishment.

   When a Token request is received, the server creates a Token for this
   particular client and sends it back to the client.

   Once a Token is retrieved from a particular server, it can be used
   for all the unicast sessions the client will be running with this
   particular server until the Token expires.  By default, Tokens are
   server specific.  However, the client can use the same Token to
   communicate with different servers if these servers are provided with
   the same secret key and algorithm used to generate the Token and are
   at least loosely clock-synchronized.

   The Token becomes invalid if the client's IP address (as seen by the
   server) changes (note that the client cannot necessarily detect this
   in a timely manner) or if the server expires the Token.  In these
   cases, the client has to request a new Token.

   As the second step, when the client wants to establish a unicast
   session, the client includes the Token with its RTCP feedback
   message.  The server validates the Token, making sure that the IP
   address information matches.  This is effective against DoS attacks,
   e.g., an attacker cannot simply spoof another client's IP address and
   start a unicast transmission towards random clients.  If the
   validation passes, the unicast session gets established.  Otherwise,
   the server notifies the client that the validation has failed, and in
   this case, the unicast session will not be established.

   Upon successful validation and once the unicast session is
   established, all the RTP and RTCP rules specified in [RFC3550] and
   other relevant specifications also apply in this session until it is
   terminated.  During the lifetime of a unicast session, a client might
   need to send RTCP messages that require authorization.  Since such
   messages require a valid Token for authorization, the client needs to
   include the Token along with such RTCP messages as explained in
   detail in later sections of this document.




Begen, et al.                Standards Track                    [Page 5]

RFC 6284                      Port Mapping                     June 2011


   Below, we first present a motivating scenario for port mapping and
   then describe the normative behavior and requirements.

3.1.  Motivating Scenario



   Consider an SSM distribution network where a distribution source
   multicasts RTP packets to a large number of clients, and one or more
   retransmission servers function as feedback targets to collect
   unicast RTCP feedback from these clients [RFC5760].  The
   retransmission servers also join the multicast session to receive the
   multicast packets and cache them for a certain time period.  When a
   client detects missing packets in the multicast session, it requests
   a retransmission from one of the retransmission servers by using an
   RTCP NACK message [RFC4585].  The retransmission server pulls the
   requested packet(s) out of the cache and retransmits them to the
   requesting client [RETRANSMISSION-FOR-SSM].

   The RTP and RTCP flows pertaining to the scenario described above are
   illustrated in Figure 2.  Between the client and server, we assume
   there exists at least one NAT device [RFC4787].  (If there are no NAT
   devices between the server and client, the method still works in the
   same fashion.)  The multicast and unicast sessions are clearly
   identified with their individual RTP and RTCP flows and port numbers.




























Begen, et al.                Standards Track                    [Page 6]

RFC 6284                      Port Mapping                     June 2011


     --------------                                 ---     ----------
    |              |-------------------------------|   |-->|P1        |
    |              |-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-|   |.->|P2        |
    |              |                               |   |   |          |
    | Distribution |      ----------------         |   |   |          |
    |    Source    |     |                |        |   |   |          |
    |              |---->|P1              |        |   |   |          |
    |              |.-.->|P2              |        |   |   |          |
    |              |     |                |        |   |   |          |
     --------------      |              P3|<.=.=.=.|   |=.=|*c0       |
                         |              P3|<~~~~~~~|   |~~~|*c1       |
    MULTICAST RTP        |                |        |   |   |          |
    SESSION with         |                |        | N |   |          |
    UNICAST FEEDBACK     |                |        | A |   |          |
                         | Retransmission |        | T |   |  Client  |
    - - - - - - - - - - -| - - - - - - - -| - - - -| - |- -| - - - - -|-
                         |     Server     |        |   |   |          |
                         |                |        |   |   |          |
    PORT MAPPING         |              PT|<~~~~~~~|   |~~>|*cT       |
                         |                |        |   |   |          |
    - - - - - - - - - - -| - - - - - - - -| - - - -| - |- -| - - - - -|-
                         |                |        |   |   |          |
    AUXILIARY UNICAST    |                |        |   |   |          |
    RTP SESSION          |                |        |   |   |          |
                         |              P3|........|   |..>|*c1       |
                         |              P3|=.=.=.=.|   |=.>|*c1       |
                         |              P4|<.=.=.=.|   |=.=|*c2       |
                         |                |        |   |   |          |
                          ----------------          ---     ----------


    -------> Multicast RTP Flow
    .-.-.-.> Multicast RTCP Flow
    .=.=.=.> Unicast RTCP Reports
    ~~~~~~~> Unicast RTCP (Feedback) Messages
    .......> Unicast RTP Flow

    Figure 2: Example Scenario Showing an SSM Distribution with Support
                     for Retransmissions from a Server

   In Figure 2, we have the following multicast and unicast ports:

   o  Ports P1 and P2 denote the destination RTP and RTCP ports in the
      multicast session, respectively.  The clients listen to these
      ports to receive the multicast RTP and RTCP packets.  Ports P1 and
      P2 are defined declaratively.





Begen, et al.                Standards Track                    [Page 7]

RFC 6284                      Port Mapping                     June 2011


   o  Port P3 denotes the RTCP port on the feedback target running on
      the retransmission server to collect any RTCP packet sent by the
      clients, including feedback messages and RTCP receiver and
      extended reports.  This is also the port that the retransmission
      server uses to send the RTP packets and RTCP sender reports in the
      unicast session.  Port P3 is defined declaratively.

   o  Port P4 denotes the RTCP port on the retransmission server used to
      collect the RTCP receiver and extended reports for the unicast
      session.  Port P4 is defined declaratively.

   o  Ports *c0, *c1, and *c2 are chosen by the client.  (Note: "*"
      indicates that the port can be chosen randomly; once chosen, the
      "*" is no longer used.) *c0 denotes the port on the client used to
      send the RTCP reports for the multicast session. *c1 denotes the
      port on the client used to send the unicast RTCP feedback messages
      in the multicast session and to receive the RTP packets and RTCP
      sender reports in the unicast session. *c2 denotes the port on the
      client used to send the RTCP receiver and extended reports in the
      unicast session.  Ports c0, c1, and c2 could be the same port or
      different ports.  There are two advantages of using the same port
      for both c0 and c1:

      1.  Some NATs only keep bindings active when a packet goes from
          the inside to the outside of the NAT (see REQ-6 of Section 4.3
          of [RFC4787]).  When the gap between the packets sent from the
          client to the server is long, this can exceed the timeout
          limit.  If c0=c1, the occasional (periodic) RTCP receiver
          reports sent from port c0 (for the multicast session's RTCP
          port P3) will ensure the NAT does not time out the public port
          associated with the incoming unicast traffic to port c1.

      2.  Having c0=c1 conserves NAT port bindings.

   o  Ports PT and *cT denote the ports through which the Token request
      and retrieval occur at the server and client sides, respectively.
      Port PT is declared on a per-unicast-session basis, although the
      same port could be used for two or more unicast sessions sourced
      by the server.  A Token once requested and retrieved by a client
      from port PT remains valid until its expiration time.

   We assume that the information declaratively defined is available as
   part of the session description information and is provided to the
   clients.  The Session Description Protocol (SDP) [RFC4566] and other
   session description methods can be used for this purpose.






Begen, et al.                Standards Track                    [Page 8]

RFC 6284                      Port Mapping                     June 2011


3.2.  Normative Behavior and Requirements



   In this section, we describe the normative behavior and requirements.
   To simplify the presentation, we refer to the port numbers described
   in the example presented in Figure 2.  However, the behavior and
   requirements described here are not specific to that particular
   example and can be applied to any scenario where analogous ports can
   be identified.

   First of all, a client compliant with this specification MUST be able
   to include a Token with any type of RTCP message (as described below)
   when it is needed.

   Second, the solution provided in this specification is not applicable
   in cases where there is RTP traffic flowing from the client to the
   server in the unicast session.  In other words, the direction of RTP
   traffic MUST be only from the server to the client in the unicast
   session.  If the client wants to send RTP traffic back to the server,
   the regular session establishment methods such as [RFC3264] need to
   be used.

   The following steps summarize the Token-based solution:

   1.  The client ascertains server address and port numbers (P3, P4 and
       PT) from the session description.  Port P4 MUST be different from
       port P3.  Port PT MAY be equal to port P3.

   2.  The client selects its local port numbers (*c0, *c1, *c2 and
       *cT).  It is strongly RECOMMENDED that the client uses the same
       port for c0 and c1.  Port cT MAY be equal to ports c0 and c1.

   3.  If the client does not have a Token (or the existing Token has
       expired):

       A.  The client first sends a Port Mapping Request message
           (Section 4.1) to port PT.  This message is sent from port cT
           on the client side.  The server learns the client's IP
           address from the received message.  The client can send this
           message anytime it wants (e.g., during initialization) and
           does not normally ever need to resend this message (see
           Section 6).

       B.  The server generates an opaque encapsulation (i.e., the
           Token) based on certain information, including the client's
           IP address.








Begen, et al.                Standards Track                    [Page 9]

RFC 6284                      Port Mapping                     June 2011


       C.  The server sends the Token back to the client using a Port
           Mapping Response message (Section 4.2).  This message MUST be
           sent from port PT towards port cT.

   4.  The client needs to provide the Token to the server using a Token
       Verification Request message (Section 4.3) whenever the client
       sends an RTCP feedback message for triggering or controlling a
       unicast session (see Section 4.3.1).  If the Token is invalid or
       missing, the server sends a Token Verification Failure message
       (Section 4.4) to the client.

       Note that the unicast session is only established after the
       server has received a feedback message (along with a valid Token)
       from the client for which it needs to react by sending unicast
       data.  Until a unicast session is established, neither the server
       nor the client needs to send RTCP reports for the unicast
       session.

   5.  Normal flows ensue as shown in Figure 2.  It is strongly
       RECOMMENDED that the client uses the same port for both c0 and
       c1, as this causes the periodic RTCP reports to keep the NAT
       mapping alive.  However, if the client uses different ports for
       c0 and c1, the client MUST keep its own NAT mapping alive for the
       P3->c1 session (see [RFC6263] for additional information).

       In the unicast session, traffic from the server to the client
       (i.e., both the RTP and RTCP packets sent from port P3 towards
       port c1) MUST be multiplexed on the same port [RFC5761].

       The client sends the RTCP receiver and extended reports in the
       unicast session from port c2 towards port P4.  The server
       correlates these reports with the reports received in the
       multicast session based on the client's RTCP Canonical Name
       (CNAME).  Thus, the client MUST use the same RTCP CNAME in both
       sessions, and its RTCP CNAME MUST be unique [RFC6222].

   A unicast session on a particular receive port c1 can last as long as
   the associated multicast session lasts.  However, a client cannot
   keep using the same receive port c1 for different subsequent unicast
   sessions since there could be packet leakage when switching from one
   unicast session to another unless each received unicast stream has
   its own distinct Synchronization Source (SSRC) identifier to allow
   the client to filter out the undesired packets.  Unless this is
   guaranteed (which is not often easy), a client SHOULD use separate
   receive ports for subsequent unicast sessions.  After a sufficient
   time (two minutes is RECOMMENDED, similar to one TCP Maximum Segment
   Lifetime specified in [RFC0793]), a previously used receive port can
   be used again.



Begen, et al.                Standards Track                   [Page 10]

RFC 6284                      Port Mapping                     June 2011


   The established unicast session can be explicitly terminated by the
   procedures specified by an application or extension using the port
   mapping approach described in this document.  In addition, the
   unicast session can also be terminated by the procedure defined
   below, which is based on timing all participants out following the
   timeout rules of [RFC3550].  Both the server and client periodically
   check the liveness of the other peer, and if there is no RTCP traffic
   from the other peer for a certain amount of time (Section 6.3.5 of
   [RFC3550] suggests five RTCP reporting intervals), the unicast
   session SHOULD be considered terminated, and no further RTP and/or
   RTCP packets SHOULD be sent in that session.  The client can attempt
   to establish a new unicast session if needed.  If no explicit
   procedure for session termination exists, the client MAY stop sending
   RTCP to the server to accomplish session termination.  However, the
   server SHALL NOT stop sending RTCP until the unicast session is
   terminated.  If Token-based authentication is also signaled to be
   allowed in the unicast session, i.e., in the RTCP messages sent from
   port c2 towards port P4, the client SHOULD terminate the unicast
   session by sending an RTCP BYE message for each SSRC it has used in
   that unicast session.

4.  Message Formats



   This section defines the formats of the RTCP messages that are
   exchanged between a server and a client for the purpose of port
   mapping.  A new RTCP control packet type is introduced, and four port
   mapping messages using this control packet are defined:

   1.  Port Mapping Request

   2.  Port Mapping Response

   3.  Token Verification Request

   4.  Token Verification Failure

   Each message has a fixed-length field for version (V), padding (P),
   sub-message type (SMT), packet type (PT), length, and SSRC of packet
   sender.  Messages have other fields as defined below.  In all
   messages defined in this section, the PT field is set to TOKEN (210).
   Individual messages are identified by the SMT field.  The length
   field indicates the message size in 32-bit words minus one, including
   the header and any padding.  This definition is in line with the
   definition of the Length field used in RTCP sender and receiver
   reports.  In all messages, any Reserved field SHALL be set to zero
   and ignored.





Begen, et al.                Standards Track                   [Page 11]

RFC 6284                      Port Mapping                     June 2011


   Following the rules specified in [RFC3550], all integer fields in the
   messages defined below are carried in network-byte order, that is,
   most significant byte (octet) first, also known as big-endian.
   Unless otherwise stated, numeric constants are in decimal (base 10).

   Note that RTCP is not a timely or reliable protocol.  The RTCP
   packets might get lost or reordered in the network, and it is not
   easy to detect these events.  When sending a new Port Mapping Request
   message, the scheduling rules that apply to sending initial RTCP
   messages [RFC4585] apply.  When a client sends a Port Mapping Request
   or Token Verification Request message but it does not receive a
   response back from the server (either a Port Mapping Response or
   Token Verification Failure message), it MAY resend its request by
   following the timer rules defined for RTCP feedback messages in
   Section 3.5 of [RFC4585] as a good practice.  However,
   implementations are advised to avoid sending spurious RTCP messages
   just because the timer rules (based on some RTCP configuration
   parameters) allow.  Reasonably safe practices are to be used to
   detect RTCP message loss.  When sending an RTCP (feedback) message
   bundled with a Token Verification Request message, the timer rules of
   [RFC4585] apply as usual.

4.1.  Port Mapping Request



   The Port Mapping Request message is identified by SMT=1.  This
   message is transmitted by the client to a dedicated server port (and
   possibly a dedicated address) to request a Token.  In the Port
   Mapping Request message, the packet sender's SSRC is set to the
   client's SSRC, which is chosen randomly by the client.  The packet
   format has the structure depicted in Figure 3.

      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|  SMT=1  |    PT=TOKEN   |         Length=3              |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                      SSRC of Packet Sender                    |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                             Random                            |
     |                             Nonce                             |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

       Figure 3: Packet Format for the Port Mapping Request Message

   o  Random Nonce (64 bits): Field that contains a random value
      generated by the client following the procedures of [RFC4086].
      This nonce is taken into account by the server when generating a
      Token for the client to enable better security for clients that



Begen, et al.                Standards Track                   [Page 12]

RFC 6284                      Port Mapping                     June 2011


      share the same IP address (such clients need to produce a random
      value extremely unlikely to collide with other clients sharing the
      same IP address).  If the same Port Mapping Request message is
      transmitted multiple times for redundancy reasons, the random
      nonce value MUST remain the same in these duplicated messages.
      However, the client MUST generate a new random nonce for every new
      Port Mapping Request message.

4.2.  Port Mapping Response



   The Port Mapping Response message is identified by SMT=2.  This
   message is sent by the server and delivers the Token to the client as
   a response to the Port Mapping Request message.  In the Port Mapping
   Response message, the packet sender's SSRC is set to the server's
   SSRC.  The packet format has the structure depicted in Figure 4.

      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|  SMT=2  |    PT=TOKEN   |          Length               |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                      SSRC of Packet Sender                    |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                    SSRC of Requesting Client                  |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                           Associated                          |
     |                             Nonce                             |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     :                         Token Element                         :
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                            Absolute                           |
     |                         Expiration Time                       |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                     Relative Expiration Time                  |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     :                       Packet Types Element                    :
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

       Figure 4: Packet Format for the Port Mapping Response Message

   o  SSRC of Requesting Client (32 bits): Field that contains the SSRC
      of the client who sent the request.

   o  Associated Nonce (64 bits): Field that contains the nonce received
      in the Port Mapping Request message and used in Token
      construction.





Begen, et al.                Standards Track                   [Page 13]

RFC 6284                      Port Mapping                     June 2011


   o  Token Element (variable size): Element that is used to carry the
      Token generated by the server.  This element is a 32-bit aligned
      Length-Value element.  The Length field, which is 16 bits,
      indicates the length (in octets) of the Value field that follows
      the Length field.  While a 16-bit length allows for Tokens with a
      size of up to 65535 bytes, using Tokens of sizes that make the
      RTCP compound packet larger than the MTU might have a negative
      impact on functionality because of IP fragmentation.  Some NATs or
      other middleboxes do not pass IP fragments; thus, a large Token
      can cause the whole mechanism to fail.  In addition, fragmentation
      increases the risk for packet loss.

      The length does not include any padding that is required for
      alignment.  The Value field carries the Token (or more accurately,
      the output of the encoding process on the server).  If the Token
      element does not fall on a 32-bit boundary, the last word MUST be
      padded to the boundary using further bits set to zero.

   o  Absolute Expiration Time (64 bits): Field that contains the
      absolute expiration time of the Token.  The absolute expiration
      time is expressed as a Network Time Protocol (NTP) timestamp value
      in seconds since the year 1900 [RFC5905].  The client does not
      need to use this element directly and thus does not need to
      synchronize its clock with the server.  However, the client needs
      to send this element back to the server along with the associated
      nonce in the Token Verification Request message and thus needs to
      keep it associated with the Token.

   o  Relative Expiration Time (32 bits): Field that contains the
      relative expiration time of the Token.  The relative expiration
      time is expressed in seconds from the time the Token was
      generated.  Whenever a server decides to not grant a Token to a
      requesting client, the relative expiration time will be set to
      zero (and hence, the accompanying Token will be invalid).

      The server conveys the relative expiration time in the clear to
      the client to allow the client to request a new Token well before
      the expiration time.

   o  Packet Types Element (variable size): Element that is used to
      signal which RTCP packet types require Token-based authentication.
      This element is a 32-bit aligned Length-Value element.  The Length
      field, which is 8 bits, indicates the length (in octets) of the
      Value field that follows the Length field.  This length does not
      include any padding that is required for alignment.  The Value
      field carries zero or more 8-bit sub-fields, each carrying an RTCP
      packet type.  If the Packet Types element does not fall on a




Begen, et al.                Standards Track                   [Page 14]

RFC 6284                      Port Mapping                     June 2011


      32-bit boundary, the last word MUST be padded to the boundary
      using further bits set to zero.  An example Packet Types element
      is shown in Figure 5.

      A server MAY change its policy on which RTCP packet types would
      require Token-based authentication based on observations,
      configuration, or other policies.  However, upon such a change,
      the server SHALL NOT send a new Port Mapping Response message to
      the clients who requested a Token earlier.  A client learns about
      this change when and if it gets a Token Verification Failure
      message.

      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
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |    Length=4   |      205      |      206      |      203      |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |      204      |                  Padding                      |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                    Figure 5: Example Packet Types Element

4.3.  Token Verification Request



   The Token Verification Request message is identified by SMT=3.  This
   message contains the Token and accompanies any RTCP message that
   would trigger a new unicast session or control an existing unicast
   session.  For a list of such messages, see Section 4.3.1.

   In the Token Verification Request message, the packet sender's SSRC
   is set to the client's SSRC.  The client MUST NOT send a Token
   Verification Request message with a Token that has expired.  The
   packet format has the structure depicted in Figure 6.


















Begen, et al.                Standards Track                   [Page 15]

RFC 6284                      Port Mapping                     June 2011


      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|  SMT=3  |    PT=TOKEN   |          Length               |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                      SSRC of Packet Sender                    |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                           Associated                          |
     |                             Nonce                             |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     :                         Token Element                         :
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                       Associated Absolute                     |
     |                         Expiration Time                       |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

    Figure 6: Packet Format for the Token Verification Request Message

   o  Associated Nonce (64 bits): Field that contains the nonce
      associated with the Token below.

   o  Token Element (variable size): Token element that was previously
      received in the Port Mapping Response message.

   o  Associated Absolute Expiration Time (64 bits): Field that contains
      the absolute expiration time associated with the Token above.

4.3.1.  Where to Include Token



   This section provides guidelines about which RTCP packet types would
   need to be accompanied by a Token Verification Request message.
   However, since a server might determine in real time that other RTCP
   messages also need to be authenticated by a Token, a client MUST act
   according to the up-to-date list provided to the client in the Port
   Mapping Response message (in the Packet Types element).  Clients need
   to support the use of Token-based authentication with any necessary
   RTCP message (see Section 3.2).

   As a general rule, when the Token capability is declared in the
   session description, the RTCP messages that trigger transmission of
   RTP packets in a port mapped unicast session are REQUIRED to be
   authenticated by using a Token.  Such messages include but are not
   limited to:

   o  NACK messages [RFC4585]

   o  RAMS Request (RAMS-R) messages [RFC6285]




Begen, et al.                Standards Track                   [Page 16]

RFC 6284                      Port Mapping                     June 2011


   Additionally, some RTCP messages might directly or indirectly control
   an existing unicast session associated with a multicast session.
   Unless another authentication method as described in their respective
   specifications is used, implementations MUST support authenticating
   such RTCP messages by using a Token.

   Examples are:

   o  BYE messages [RFC3550]

   o  RAMS Termination (RAMS-T) messages [RFC6285]

   o  Codec Control Messages (CCMs) [RFC5104]

   Note that even if a packet type is listed to require Token-based
   authentication, it does not need to be authenticated when it does not
   control the unicast session.  For example, if BYE (203) is listed in
   the Port Mapping Response message as one of the packet types that
   requires authentication, the client does not need to bundle the RTCP
   BYE message with a Token when it is sending it for the multicast
   session.

   The Token Verification Request message might also be bundled with
   packets carrying RTCP receiver and/or extended reports.  While such
   packets do not have a strong security impact, a specific application
   might desire to have a more controlled reporting scheme from the
   clients.  In this case, the server lists the packet types for the
   receiver (201) and/or extended reports (207) in the Port Mapping
   Response message.

4.4.  Token Verification Failure



   The Token Verification Failure message is identified by SMT=4.  This
   message is sent by the server and notifies the client that the Token
   was invalid or that the client did not include a Token Verification
   Request message in the RTCP packet although it was supposed to (the
   message is sent from port P3 towards port c1).  The packet format has
   the structure depicted in Figure 7.













Begen, et al.                Standards Track                   [Page 17]

RFC 6284                      Port Mapping                     June 2011


      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|  SMT=4  |    PT=TOKEN   |         Length=5              |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                      SSRC of Packet Sender                    |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                    SSRC of Requesting Client                  |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |   Failed PT   |   FMT   |              Reserved               |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                           Associated                          |
     |                             Nonce                             |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

    Figure 7: Packet Format for the Token Verification Failure Message

   o  SSRC of Packet Sender: This is the server's SSRC, which equals the
      SSRC of the respective multicast stream.  Note that this SSRC
      value is from a different SSRC space than the one used in the
      unicast session.

   o  SSRC of Requesting Client (32 bits): Field that contains the SSRC
      of the client.

   o  Failed PT (8 bits): Field that indicates the type of the RTCP
      packet that caused this failure message.

   o  FMT (5 bits): Field that indicates the feedback message type (FMT)
      value of the RTCP packet that caused this failure.  Together with
      the field above, the client can infer which RTCP message it had
      previously sent caused this failure message to be sent by the
      server.  For example, if the client did not include a valid Token
      with an RTCP NACK message, the Failed PT field will indicate 205
      (RTPFB) and the FMT field will indicate 1 (Generic NACK).  If the
      RTCP message did not have an associated FMT value (such as an RTCP
      BYE message), the FMT field SHALL be set to zero.

   o  Associated Nonce (64 bits): Field that contains the nonce received
      in the Token Verification Request message.  If there was no Token
      Verification Request message included by the client, this field is
      set to zero.

5.  Procedures for Token Construction



   The Token encoding is known to the server but opaque to the client.
   Implementations MUST encode the following information into the Token
   as a minimum, in order to provide adequate security:



Begen, et al.                Standards Track                   [Page 18]

RFC 6284                      Port Mapping                     June 2011


   o  Client's IP address as seen by the server (32/128 bits for IPv4/
      IPv6 addresses)

   o  The nonce generated and inserted in the Port Mapping Request
      message by the client (64 bits)

   o  The absolute expiration time chosen by the server indicated as an
      NTP timestamp value in seconds since the year 1900 [RFC5905] (64
      bits, to protect against replay attacks)

   The RECOMMENDED way for constructing Tokens is to perform HMAC-SHA1
   [RFC2104] on the concatenated values of the information listed above
   (implementations might adopt different approaches).  If HMAC-SHA1 is
   used, the Hashed Message Authentication Code (HMAC) key MUST be at
   least 160 bits long and generated using a cryptographically secure
   random source [RFC4086].

   In addition to the information listed above, implementations are
   encouraged to encode whatever additional information is deemed
   necessary or useful.  For example, key rollover is simplified by
   encoding a key-id into the Token.  As another example, a cluster of
   anycast servers could find advantage by encoding a server identifier
   into the Token.  As another example, while HMAC-SHA1 provides a level
   of security that is widely regarded as being more than sufficient for
   providing message authentication and it is secure against all known
   cryptanalytic attacks that use computational resources that are
   currently economically feasible, a replacement HMAC algorithm (e.g.,
   HMAC-SHA256) could be used instead if HMAC-SHA1 has been compromised.

   To protect from offline attacks, the server SHOULD occasionally
   choose a new HMAC key.  To ease implementation, a key-id can be
   assigned to each HMAC key.  This can be encoded as simply as one bit
   (where the new key is X (e.g., 1) and the old key is the inverted
   value of X (e.g., 0)), or if several keys are supported at once, the
   key-id could be encoded into several bits.  As the encoding of the
   Token is entirely private to the server and opaque to the clients,
   any encoding can be used.  By encoding the key-id into the Token
   element, the server can reject an old key without bothering to do
   HMAC validation (saving CPU cycles).  The key-id can be encoded into
   the Value field of the Token element by simply concatenating the
   (plaintext) key-id with the hashed information (i.e., the Token
   itself).

   For example, the Value field in the Token element can be computed as:

           key-id || mac-alg (client-ip | nonce | abs-expiration)





Begen, et al.                Standards Track                   [Page 19]

RFC 6284                      Port Mapping                     June 2011


   During Token construction, the expiration time has to be chosen
   carefully based on the intended service duration.  Tokens that are
   valid for an unnecessarily long period of time (e.g., several hours)
   might impose security risks.  Depending on the application and use
   cases, a reasonable value needs to be chosen by the server.  Note
   that using shorter lifetimes requires the clients to acquire Tokens
   more frequently.  However, since a client can acquire a new Token
   well before it will need to use it, the client will not necessarily
   be penalized for the acquisition delay.

   Finally, be aware that NTP timestamps will wrap around in the year
   2036.  Refer to Section 6 of [RFC5905] for further details.

6.  Validating Tokens



   The server MUST validate the Token upon receipt of an RTCP feedback
   message along with the Token Verification Request message that
   contains a Token, nonce, and absolute expiration time.

   The server first applies its own procedure for constructing the
   Tokens by using the client's IP address from the received Token
   Verification Request message and the nonce and absolute expiration
   time values reported in the received Token Verification Request
   message.  The server then compares the resulting output with the
   Token sent by the client in the Token Verification Request message.
   If they match and the absolute expiration time has not passed yet,
   the server declares that the Token is valid.

   Note that if the client's IP address changes, the Token will not
   validate.  Similarly, if the client inserts an incorrect nonce or
   absolute expiration time value in the Token Verification Request
   message, validation will fail.  It is also possible that the server
   wants to expire the Token prematurely.  In these cases, the server
   MUST reply back to the client with a Token Verification Failure
   message (that goes from port P3 on the server towards port c1 on the
   client).

   In addition to the Token Verification Failure message, it is
   RECOMMENDED that applications define an application-specific error
   response to be sent by the server when the server detects that the
   Token is invalid.  For applications using [RFC6285], this document
   defines a new 4xx-level response code in the RAMS Response Code Space
   Registry.  A client that receives a Token Verification Failure
   message can request a new Token from the server.

   If a client receives a Port Mapping Response message with an invalid
   Token (i.e., the relative expiration time is set to zero) two or more
   times for a particular Port Mapping Request message or the client



Begen, et al.                Standards Track                   [Page 20]

RFC 6284                      Port Mapping                     June 2011


   receives a Token Verification Failure message two or more times for
   the same Token Verification Request message, the client SHOULD do the
   following:

   1.  Check whether or not the session description has been updated.
       If it was updated, act according to the new session description.

   2.  Exponentially back off for the third and subsequent attempts.
       Exponential back-off does not apply when the client sends a Port
       Mapping Request or Token Verification Request message to a new
       address and/or port.

7.  SDP Signaling



7.1.  The 'portmapping-req' Attribute



   This attribute is used declaratively in any media block that
   describes an RTP session that uses Token-based authentication for one
   or more RTCP messages relating to that session.  It indicates the
   port and optionally the address for obtaining a Token.

   The presence of the 'portmapping-req' attribute indicates that (i) a
   Token MUST be included in certain RTCP messages sent to the server
   triggering or controlling a unicast session (see Section 4.3.1) and
   (ii) the client MUST receive the unicast session's RTP and RTCP
   packets from the server on the port from which it sent the RTCP
   message triggering the unicast session.

      Note: This does not imply that Token Verification Request messages
      always need to be sent in the unicast session.  Token Verification
      Request messages accompany RTCP messages that trigger or control
      this unicast session and are sent either in the multicast session
      or the unicast session, depending on the RTCP message (see
      Section 4.3.1).

7.1.1.  ABNF Definition of 'portmapping-req'



   The formal description of the 'portmapping-req' attribute is defined
   by the following ABNF [RFC5234] syntax:

      portmapping-req-attr = "a=portmapping-req:" port [SP nettype SP
                               addrtype SP connection-address] CRLF

   Here, 'port', 'nettype', 'addrtype', and 'connection-address' are
   defined as specified in Section 9 of [RFC4566].

   The 'portmapping-req' attribute SHALL only be used as a media-level
   attribute.



Begen, et al.                Standards Track                   [Page 21]

RFC 6284                      Port Mapping                     June 2011


   In the optional address value, only unicast addresses SHOULD be used
   unless one wants to use a multicast address after evaluating the
   additional security risks such as non-legit servers generating fake
   Tokens.  If the address is not specified, the (source) address in the
   "c" line applicable to the media description SHALL be used.

7.1.2.  Offer/Answer Model Considerations



   When using the 'portmapping-req' attribute in SDP offer/answer
   exchanges [RFC3264], the following considerations apply.  When an
   offerer sends an answerer an offer of an SDP description making use
   of the Token approach described in this specification, the
   'portmapping-req' attribute is included declaratively.  There will
   not be offer/answer exchanges between the answerer and the actual
   server providing the unicast service(s).

   When the answerer supports the Token approach, it MUST echo in its
   answer back to the offerer the 'portmapping-req' attribute from the
   offer including the same port number and address (if any).  If the
   answerer does not implement this specification, it follows normal SDP
   parsing of unknown attributes (they are ignored and are not sent in
   the answer).  This means that the answerer can still join the
   multicast session but will not be able to use the unicast service(s)
   that require the use of Tokens.

7.2.  Requirements



   The use of SDP for the port mapping solution normatively requires
   support for:

   o  The SDP grouping framework and flow identification (FID) semantics
      [RFC5888]

   o  The RTP/Audio-Visual Profile with Feedback (AVPF) profile
      [RFC4585]

   o  The 'rtcp-mux' attribute (to multiplex RTP and RTCP on a single
      port on both endpoints in the unicast session [RFC5761])













Begen, et al.                Standards Track                   [Page 22]

RFC 6284                      Port Mapping                     June 2011


7.3.  Example and Discussion



   The declarative SDP describing the scenario given in Figure 2 is
   written as:

        v=0
        o=ali 1122334455 1122334466 IN IP4 nack.example.com
        s=Local Retransmissions
        t=0 0
        a=group:FID 1 2
        a=rtcp-unicast:rsi
        m=video 41000 RTP/AVPF 98
        i=Multicast Stream
        c=IN IP4 233.252.0.2/255
        a=source-filter:incl IN IP4 233.252.0.2 198.51.100.1   ; Note 1
        a=rtpmap:98 MP2T/90000
        a=multicast-rtcp:41500                                 ; Note 1
        a=rtcp:42000 IN IP4 192.0.2.1                          ; Note 2
        a=rtcp-fb:98 nack                                      ; Note 2
        a=portmapping-req:30000 IN IP4 192.0.2.1               ; Note 3
        a=mid:1
        m=video 42000 RTP/AVPF 99                              ; Note 4
        i=Unicast Retransmission Stream
        c=IN IP4 192.0.2.1
        a=sendonly
        a=rtpmap:99 rtx/90000
        a=rtcp-mux                                             ; Note 5
        a=rtcp:42500                                           ; Note 6
        a=fmtp:99 apt=98; rtx-time=5000
        a=portmapping-req:30001                                ; Note 3
        a=mid:2

       Figure 8: SDP Describing an SSM Distribution with Support for
                    Retransmissions from a Local Server

   In this description, we highlight the following notes:

   Note 1: The source stream is multicast from a distribution source
   with a source IP address of 198.51.100.1 to the multicast destination
   address of 233.252.0.2 and port 41000 (P1).  The associated RTCP
   packets are multicast in the same group to port 41500 (P2).

   Note 2: A retransmission server including feedback target
   functionality with an IP address of 192.0.2.1 and port of 42000 (P3)
   is specified with the 'rtcp' attribute.  The feedback functionality
   is enabled for the RTP stream with payload type 98 through the
   'rtcp-fb' attribute [RFC4585].




Begen, et al.                Standards Track                   [Page 23]

RFC 6284                      Port Mapping                     June 2011


   Note 3: The "a=portmapping-req" line indicates that one or more RTCP
   messages relating to the RTP session described in this media block
   uses Token-based authentication, and a Token needs to be retrieved
   first from the designated port (PT) before the unicast session can be
   established.  In the first appearance, an explicit address is
   provided.  In the second appearance, there is no address indicated in
   this line and the client needs to send the Token request to the
   address specified in the "c" line in the unicast media block.

   Note 4: The port specified in the second "m" line (for the unicast
   stream) does not mean anything in this scenario as the client does
   not send any RTP traffic back to the server.

   Note 5: The server multiplexes RTP and RTCP packets sent towards c1
   on the same port.

   Note 6: The server uses port 42500 (P4) for the unicast session.

8.  Address Pooling NATs



   Large-scale NAT devices have a pool of public IPv4 addresses and map
   internal hosts to one of those public IPv4 addresses.  As long as an
   internal host maintains an active mapping in the NAT, the same IPv4
   address is assigned to new connections.  However, once all of the
   host's mappings have been deleted (e.g., because of timeout), it is
   possible that a new connection from that same host will be assigned a
   different IPv4 address from the pool.  When that occurs, the Token
   will be considered invalid by the server, causing an additional round
   trip for the client to acquire a fresh Token.

   Any traffic from the host that traverses the NAT will prevent this
   problem.  As the host is sending RTCP receiver reports at least every
   5 seconds (Section 6.2 of [RFC3550]) for the multicast session it is
   receiving, those RTCP messages will be sufficient to prevent this
   problem.

9.  Security Considerations



9.1.  Tokens



   The Token, which is generated based on a client's IP address and
   expiration date, provides protection against off-path denial-of-
   service (DoS) attacks.  An attacker using a certain IP address cannot
   cause one or more RTP packets to be sent to a victim client who has a
   different IP address.  However, if the attacker acquires a valid
   Token for a victim and can spoof the victim's source address, this





Begen, et al.                Standards Track                   [Page 24]

RFC 6284                      Port Mapping                     June 2011


   approach becomes vulnerable to replay attacks.  This is especially
   easy if the attacker and victim are behind a large-scale NAT and
   share the same IP address.

   Multicast is deployed on managed networks, not the Internet.  These
   managed networks will choose whether or not to enable network ingress
   filtering [RFC2827].  If ingress filtering is enabled on a network,
   an attacker cannot spoof a victim's IP address to use a Token to
   initiate an attack against a victim.  However, if ingress filtering
   is not enabled on a network, an attacker could obtain a Token and
   spoof the victim's address, causing traffic to flood the victim.  On
   such a network, the server can reduce the time period for such an
   attack by expiring a Token in a short period of time.  In the extreme
   case, the server can expire the Token in such a short period of time
   that the client will have to acquire a new Token immediately before
   using it in a Token Verification Request message.  One should,
   however, note that such a behavior might have an adverse effect on
   the delay in establishing or controlling a unicast session.

   RTCP messages could be subject to on-path or man-in-the-middle
   attacks.  For example, an attacker can modify a value in one or more
   fields in the Port Mapping Response or the Token Verification Request
   message that are used in Token construction.  This will result in
   Token validation failure.  Consequently, the client ends up asking
   the server to generate a new Token.  The resulting delay and extra
   processing on the server is undesirable.

   Alternatively, the attacker can modify a value in a field that is not
   used in Token construction.  For example, the attacker can reduce the
   value in the Relative Expiration Time field in the Port Mapping
   Response message from two hours to two minutes.  While the Token will
   still validate, this attack will result in more frequent requests to
   the server for a new Token.  Oppositely, the attacker can increase
   the value in the Relative Expiration Time field and make the client
   think the Token will be valid for a longer time.  This attack can be
   only detected by monitoring the activity on the server.  Note that
   using the relative expiration time in Token construction does not
   necessarily make this attack easier to detect since the attacker
   might revert the modified value back to its original value in the
   Token Verification Request message.  This allows the Token to still
   validate on the server.  In this case, the attack is still only
   detectable by monitoring the server activity.

   If there is a risk or concern for on-path or man-in-the-middle
   attacks, RTCP messages SHOULD be protected by Secure RTCP (SRTCP)
   [RFC3711].





Begen, et al.                Standards Track                   [Page 25]

RFC 6284                      Port Mapping                     June 2011


   To minimize the risk of cross-protocol attacks, a server MUST NOT use
   the same secret key it used for Token construction for other
   purposes.

9.2.  The 'portmapping-req' Attribute



   The 'portmapping-req' attribute is not believed to introduce any
   significant security risk to multimedia applications.  A malevolent
   third party could use this attribute to redirect the Port Mapping
   Request messages by altering the port number or cause the unicast
   session establishment to fail by removing it from the SDP
   description.  However, this requires intercepting and rewriting the
   packets carrying the SDP description, and if an interceptor can do
   that, many more attacks are possible, including a wholesale change of
   the addresses and port numbers at which the media will be sent.

   In order to avoid attacks of this sort, the SDP description needs to
   be integrity protected and provided with source authentication.  This
   can, for example, be achieved on an end-to-end basis using Secure/
   Multipurpose Internet Mail Extensions (S/MIME) [RFC5652] [RFC5751]
   when SDP is used in a signaling packet using MIME types (application/
   sdp).  Alternatively, HTTPS [RFC2818] or the authentication method in
   the Session Announcement Protocol (SAP) [RFC2974] could be used as
   well.

10.  IANA Considerations



   The following contact information is used for all registrations in
   this document:

   Ali Begen
   abegen@cisco.com

10.1.  Registration of SDP Attributes



   This document registers one new attribute name in SDP.

        SDP Attribute ("att-field"):
        Attribute name:     portmapping-req
        Long form:          Port and address for requesting Token
        Type of name:       att-field
        Type of attribute:  Media level
        Subject to charset: No
        Purpose:            See this document
        Reference:          [RFC6284]
        Values:             See this document





Begen, et al.                Standards Track                   [Page 26]

RFC 6284                      Port Mapping                     June 2011


10.2.  Registration of RTCP Control Packet Types



   In accordance with Section 15 of [RFC3550], this specification adds
   the following value to the RTCP Control Packet types sub-registry in
   the Real-Time Transport Protocol (RTP) Parameters registry:

   Value     Abbrev.    Name                                   Reference
   --------  ---------  -------------------------------------  ---------
   210       TOKEN      Port Mapping                           [RFC6284]

10.3.  SMT Values for TOKEN Packet Type Registry



   This document creates a new sub-registry for the sub-message type
   (SMT) values to be used with the TOKEN packet type.  The registry is
   called the SMT Values for TOKEN Packet Type Registry.  This registry
   is managed by the IANA according to the IETF Review policy of
   [RFC5226].

   The length of the SMT field is five bits, allowing 32 values.  The
   registry is initialized with the following entries:

   Value Name                                               Reference
   ----- -------------------------------------------------- ------------
   0     Reserved                                           [RFC6284]
   1     Port Mapping Request                               [RFC6284]
   2     Port Mapping Response                              [RFC6284]
   3     Token Verification Request                         [RFC6284]
   4     Token Verification Failure                         [RFC6284]
   5-30  Unassigned                                         IETF Review
   31    Reserved                                           [RFC6284]

   The SMT values 0 and 31 are reserved for future use.

10.4.  RAMS Response Code Space Registry



   This document adds the following entry to the RAMS Response Code
   Space Registry.

   Code  Description                                        Reference
   ----- -------------------------------------------------- ------------
   405   Invalid Token                                      [RFC6284]

   This response code is used when the Token included by the RTP_Rx in
   the RAMS-R message is invalid.







Begen, et al.                Standards Track                   [Page 27]

RFC 6284                      Port Mapping                     June 2011


11.  Acknowledgments



   The approach presented in this document came out after discussions
   with various individuals in the AVT and MMUSIC WGs and the breakout
   session held at the Anaheim meeting.  We thank each of these
   individuals, particularly Magnus Westerlund and Colin Perkins.

12.  References



12.1.  Normative References



   [RFC2104]  Krawczyk, H., Bellare, M., and R. Canetti, "HMAC: Keyed-
              Hashing for Message Authentication", RFC 2104,
              February 1997.

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119, March 1997.

   [RFC3550]  Schulzrinne, H., Casner, S., Frederick, R., and V.
              Jacobson, "RTP: A Transport Protocol for Real-Time
              Applications", STD 64, RFC 3550, July 2003.

   [RFC3711]  Baugher, M., McGrew, D., Naslund, M., Carrara, E., and K.
              Norrman, "The Secure Real-time Transport Protocol (SRTP)",
              RFC 3711, March 2004.

   [RFC4086]  Eastlake, D., Schiller, J., and S. Crocker, "Randomness
              Requirements for Security", BCP 106, RFC 4086, June 2005.

   [RFC4566]  Handley, M., Jacobson, V., and C. Perkins, "SDP: Session
              Description Protocol", RFC 4566, July 2006.

   [RFC4585]  Ott, J., Wenger, S., Sato, N., Burmeister, C., and J. Rey,
              "Extended RTP Profile for Real-time Transport Control
              Protocol (RTCP)-Based Feedback (RTP/AVPF)", RFC 4585,
              July 2006.

   [RFC5234]  Crocker, D. and P. Overell, "Augmented BNF for Syntax
              Specifications: ABNF", STD 68, RFC 5234, January 2008.

   [RFC5760]  Ott, J., Chesterfield, J., and E. Schooler, "RTP Control
              Protocol (RTCP) Extensions for Single-Source Multicast
              Sessions with Unicast Feedback", RFC 5760, February 2010.

   [RFC5761]  Perkins, C. and M. Westerlund, "Multiplexing RTP Data and
              Control Packets on a Single Port", RFC 5761, April 2010.





Begen, et al.                Standards Track                   [Page 28]

RFC 6284                      Port Mapping                     June 2011


   [RFC5888]  Camarillo, G. and H. Schulzrinne, "The Session Description
              Protocol (SDP) Grouping Framework", RFC 5888, June 2010.

   [RFC5905]  Mills, D., Martin, J., Burbank, J., and W. Kasch, "Network
              Time Protocol Version 4: Protocol and Algorithms
              Specification", RFC 5905, June 2010.

   [RFC6222]  Begen, A., Perkins, C., and D. Wing, "Guidelines for
              Choosing RTP Control Protocol (RTCP) Canonical Names
              (CNAMEs)", RFC 6222, April 2011.

12.2.  Informative References



   [RETRANSMISSION-FOR-SSM]
              Van Caenegem, T., Ver Steeg, B., and A. Begen,
              "Retransmission for Source-Specific Multicast (SSM)
              Sessions", Work in Progress, May 2011.

   [RFC0793]  Postel, J., "Transmission Control Protocol", STD 7,
              RFC 793, September 1981.

   [RFC2818]  Rescorla, E., "HTTP Over TLS", RFC 2818, May 2000.

   [RFC2827]  Ferguson, P. and D. Senie, "Network Ingress Filtering:
              Defeating Denial of Service Attacks which employ IP Source
              Address Spoofing", BCP 38, RFC 2827, May 2000.

   [RFC2974]  Handley, M., Perkins, C., and E. Whelan, "Session
              Announcement Protocol", RFC 2974, October 2000.

   [RFC3264]  Rosenberg, J. and H. Schulzrinne, "An Offer/Answer Model
              with Session Description Protocol (SDP)", RFC 3264,
              June 2002.

   [RFC4145]  Yon, D. and G. Camarillo, "TCP-Based Media Transport in
              the Session Description Protocol (SDP)", RFC 4145,
              September 2005.

   [RFC4607]  Holbrook, H. and B. Cain, "Source-Specific Multicast for
              IP", RFC 4607, August 2006.

   [RFC4787]  Audet, F. and C. Jennings, "Network Address Translation
              (NAT) Behavioral Requirements for Unicast UDP", BCP 127,
              RFC 4787, January 2007.

   [RFC5104]  Wenger, S., Chandra, U., Westerlund, M., and B. Burman,
              "Codec Control Messages in the RTP Audio-Visual Profile
              with Feedback (AVPF)", RFC 5104, February 2008.



Begen, et al.                Standards Track                   [Page 29]

RFC 6284                      Port Mapping                     June 2011


   [RFC5226]  Narten, T. and H. Alvestrand, "Guidelines for Writing an
              IANA Considerations Section in RFCs", BCP 26, RFC 5226,
              May 2008.

   [RFC5652]  Housley, R., "Cryptographic Message Syntax (CMS)", STD 70,
              RFC 5652, September 2009.

   [RFC5751]  Ramsdell, B. and S. Turner, "Secure/Multipurpose Internet
              Mail Extensions (S/MIME) Version 3.2 Message
              Specification", RFC 5751, January 2010.

   [RFC6263]  Marjou, X. and A. Sollaud, "Application Mechanism for
              Keeping Alive the NAT Mappings Associated with RTP / RTP
              Control Protocol (RTCP) Flows", RFC 6263, June 2011.

   [RFC6285]  Ver Steeg, B., Begen, A., Van Caenegem, T., and Z. Vax,
              "Unicast-Based Rapid Acquisition of Multicast RTP
              Sessions", RFC 6285, June 2011.

Authors' Addresses



   Ali Begen
   Cisco
   181 Bay Street
   Toronto, ON  M5J 2T3
   Canada

   EMail: abegen@cisco.com


   Dan Wing
   Cisco
   170 West Tasman Dr.
   San Jose, CA  95134
   USA

   EMail: dwing@cisco.com


   Tom Van Caenegem
   Alcatel-Lucent
   Copernicuslaan 50
   Antwerpen  2018
   Belgium

   EMail: Tom.Van_Caenegem@alcatel-lucent.com





Begen, et al.                Standards Track                   [Page 30]