RFC 5850

Internet Engineering Task Force (IETF)                           R. Mahy
Request for Comments: 5850                                  Unaffiliated
Category: Informational                                        R. Sparks
ISSN: 2070-1721                                                  Tekelec
                                                            J. Rosenberg
                                                               D. Petrie
                                                        A. Johnston, Ed.
                                                                May 2010

           A Call Control and Multi-Party Usage Framework for
                 the Session Initiation Protocol (SIP)


   This document defines a framework and the requirements for call
   control and multi-party usage of the Session Initiation Protocol
   (SIP).  To enable discussion of multi-party features and
   applications, we define an abstract call model for describing the
   media relationships required by many of these.  The model and actions
   described here are specifically chosen to be independent of the SIP
   signaling and/or mixing approach chosen to actually set up the media
   relationships.  In addition to its dialog manipulation aspect, this
   framework includes requirements for communicating related information
   and events such as conference and session state and session history.
   This framework also describes other goals that embody the spirit of
   SIP applications as used on the Internet such as the definition of
   primitives (not services), invoker and participant oriented
   primitives, signaling and mixing model independence, and others.

Status of This Memo

   This document is not an Internet Standards Track specification; it is
   published for informational purposes.

   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).  Not all documents
   approved by the IESG are a candidate for any level of Internet
   Standard; see 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

Mahy, et al.                  Informational                     [Page 1]

RFC 5850               SIP Call Control Framework               May 2010

Copyright Notice

   Copyright (c) 2010 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.

   This document may contain material from IETF Documents or IETF
   Contributions published or made publicly available before November
   10, 2008.  The person(s) controlling the copyright in some of this
   material may not have granted the IETF Trust the right to allow
   modifications of such material outside the IETF Standards Process.
   Without obtaining an adequate license from the person(s) controlling
   the copyright in such materials, this document may not be modified
   outside the IETF Standards Process, and derivative works of it may
   not be created outside the IETF Standards Process, except to format
   it for publication as an RFC or to translate it into languages other
   than English.

Table of Contents

   1.  Motivation and Background  . . . . . . . . . . . . . . . . . .  4
   2.  Key Concepts . . . . . . . . . . . . . . . . . . . . . . . . .  6
     2.1.  Conversation Space Model . . . . . . . . . . . . . . . . .  7
     2.2.  Relationship between Conversation Space, SIP Dialogs,
           and SIP Sessions . . . . . . . . . . . . . . . . . . . . .  8
     2.3.  Signaling Models . . . . . . . . . . . . . . . . . . . . .  9
     2.4.  Mixing Models  . . . . . . . . . . . . . . . . . . . . . . 10
       2.4.1.  Tightly Coupled  . . . . . . . . . . . . . . . . . . . 11
       2.4.2.  Loosely Coupled  . . . . . . . . . . . . . . . . . . . 12
     2.5.  Conveying Information and Events . . . . . . . . . . . . . 13
     2.6.  Componentization and Decomposition . . . . . . . . . . . . 15
       2.6.1.  Media Intermediaries . . . . . . . . . . . . . . . . . 15
       2.6.2.  Text-to-Speech and Automatic Speech Recognition  . . . 17
       2.6.3.  VoiceXML . . . . . . . . . . . . . . . . . . . . . . . 17
     2.7.  Use of URIs  . . . . . . . . . . . . . . . . . . . . . . . 18
       2.7.1.  Naming Users in SIP  . . . . . . . . . . . . . . . . . 19
       2.7.2.  Naming Services with SIP URIs  . . . . . . . . . . . . 20
     2.8.  Invoker Independence . . . . . . . . . . . . . . . . . . . 22
     2.9.  Billing Issues . . . . . . . . . . . . . . . . . . . . . . 23

Mahy, et al.                  Informational                     [Page 2]

RFC 5850               SIP Call Control Framework               May 2010

   3.  Catalog of Call Control Actions and Sample Features  . . . . . 23
     3.1.  Remote Call Control Actions on Early Dialogs . . . . . . . 24
       3.1.1.  Remote Answer  . . . . . . . . . . . . . . . . . . . . 24
       3.1.2.  Remote Forward or Put  . . . . . . . . . . . . . . . . 24
       3.1.3.  Remote Busy or Error Out . . . . . . . . . . . . . . . 24
     3.2.  Remote Call Control Actions on Single Dialogs  . . . . . . 24
       3.2.1.  Remote Dial  . . . . . . . . . . . . . . . . . . . . . 24
       3.2.2.  Remote On and Off Hold . . . . . . . . . . . . . . . . 25
       3.2.3.  Remote Hangup  . . . . . . . . . . . . . . . . . . . . 25
     3.3.  Call Control Actions on Multiple Dialogs . . . . . . . . . 25
       3.3.1.  Transfer . . . . . . . . . . . . . . . . . . . . . . . 25
       3.3.2.  Take . . . . . . . . . . . . . . . . . . . . . . . . . 26
       3.3.3.  Add  . . . . . . . . . . . . . . . . . . . . . . . . . 27
       3.3.4.  Local Join . . . . . . . . . . . . . . . . . . . . . . 28
       3.3.5.  Insert . . . . . . . . . . . . . . . . . . . . . . . . 28
       3.3.6.  Split  . . . . . . . . . . . . . . . . . . . . . . . . 29
       3.3.7.  Near-Fork  . . . . . . . . . . . . . . . . . . . . . . 29
       3.3.8.  Far-Fork . . . . . . . . . . . . . . . . . . . . . . . 29
   4.  Security Considerations  . . . . . . . . . . . . . . . . . . . 30
   Appendix A.    Example Features  . . . . . . . . . . . . . . . . . 32
   Appendix A.1.  Attended Transfer . . . . . . . . . . . . . . . . . 32
   Appendix A.2.  Auto Answer . . . . . . . . . . . . . . . . . . . . 32
   Appendix A.3.  Automatic Callback  . . . . . . . . . . . . . . . . 32
   Appendix A.4.  Barge-In  . . . . . . . . . . . . . . . . . . . . . 32
   Appendix A.5.  Blind Transfer  . . . . . . . . . . . . . . . . . . 32
   Appendix A.6.  Call Forwarding . . . . . . . . . . . . . . . . . . 33
   Appendix A.7.  Call Monitoring . . . . . . . . . . . . . . . . . . 33
   Appendix A.8.  Call Park . . . . . . . . . . . . . . . . . . . . . 33
   Appendix A.9.  Call Pickup . . . . . . . . . . . . . . . . . . . . 33
   Appendix A.10. Call Return . . . . . . . . . . . . . . . . . . . . 34
   Appendix A.11. Call Waiting  . . . . . . . . . . . . . . . . . . . 34
   Appendix A.12. Click-to-Dial . . . . . . . . . . . . . . . . . . . 34
   Appendix A.13. Conference Call . . . . . . . . . . . . . . . . . . 34
   Appendix A.14. Consultative Transfer . . . . . . . . . . . . . . . 34
   Appendix A.15. Distinctive Ring  . . . . . . . . . . . . . . . . . 35
   Appendix A.16. Do Not Disturb  . . . . . . . . . . . . . . . . . . 35
   Appendix A.17. Find-Me . . . . . . . . . . . . . . . . . . . . . . 35
   Appendix A.18. Hotline . . . . . . . . . . . . . . . . . . . . . . 35
   Appendix A.19. IM Conference Alerts  . . . . . . . . . . . . . . . 35
   Appendix A.20. Inbound Call Screening  . . . . . . . . . . . . . . 35
   Appendix A.21. Intercom  . . . . . . . . . . . . . . . . . . . . . 36
   Appendix A.22. Message Waiting . . . . . . . . . . . . . . . . . . 36
   Appendix A.23. Music on Hold . . . . . . . . . . . . . . . . . . . 36
   Appendix A.24. Outbound Call Screening . . . . . . . . . . . . . . 36
   Appendix A.25. Pre-Paid Calling  . . . . . . . . . . . . . . . . . 37
   Appendix A.26. Presence-Enabled Conferencing . . . . . . . . . . . 37
   Appendix A.27. Single Line Extension/Multiple Line Appearance  . . 37
   Appendix A.28. Speakerphone Paging . . . . . . . . . . . . . . . . 38

Mahy, et al.                  Informational                     [Page 3]

RFC 5850               SIP Call Control Framework               May 2010

   Appendix A.29. Speed Dial  . . . . . . . . . . . . . . . . . . . . 38
   Appendix A.30. Voice Message Screening . . . . . . . . . . . . . . 38
   Appendix A.31. Voice Portal  . . . . . . . . . . . . . . . . . . . 39
   Appendix A.32. Voicemail . . . . . . . . . . . . . . . . . . . . . 40
   Appendix A.33. Whispered Call Waiting  . . . . . . . . . . . . . . 40
   Appendix B.    Acknowledgments . . . . . . . . . . . . . . . . . . 40
   5.  Informative References . . . . . . . . . . . . . . . . . . . . 40

1.  Motivation and Background

   The Session Initiation Protocol (SIP) [RFC3261] was defined for the
   initiation, maintenance, and termination of sessions or calls between
   one or more users.  However, despite its origins as a large-scale
   multi-party conferencing protocol, SIP is used today primarily for
   point-to-point calls.  This two-party configuration is the focus of
   the SIP specification and most of its extensions.

   This document defines a framework and the requirements for call
   control and multi-party usage of SIP.  Most multi-party operations
   manipulate SIP dialogs (also known as call legs) or SIP conference
   media policy to cause participants in a conversation to perceive
   specific media relationships.  In other protocols that deal with the
   concept of calls, this manipulation is known as call control.  In
   addition to its dialog or policy manipulation aspect, call control
   also includes communicating information and events related to
   manipulating calls, including information and events dealing with
   session state and history, conference state, user state, and even
   message state.

   Based on input from the SIP community, the authors compiled the
   following set of goals for SIP call control and multi-party

   o  Define primitives, not services.  Allow for a handful of robust
      yet simple mechanisms that can be combined to deliver features and
      services.  Throughout this document, we refer to these simple
      mechanisms as "primitives".  Primitives should be sufficiently
      robust so that when they are combined with each other, they can be
      used to build lots of services.  However, the goal is not to
      define a provably complete set of primitives.  Note that while the
      IETF will NOT standardize behavior or services, it may define
      example services for informational purposes, as in service
      examples [RFC5359].

   o  Be participant oriented.  The primitives should be designed to
      provide services that are oriented around the experience of the
      participants.  The authors observe that end users of features and
      services usually don't care how a media relationship is set up.

Mahy, et al.                  Informational                     [Page 4]

RFC 5850               SIP Call Control Framework               May 2010

      Their ultimate experience is only based on the resulting media and
      other externally visible characteristics.

   o  Be signaling model independent.  Support both a central-control
      and a peer-to-peer feature invocation model (and combinations of
      the two).  Baseline SIP already supports a centralized control
      model described in 3pcc (third party call control) [RFC3725], and
      the SIP community has expressed a great deal of interest in peer-
      to-peer or distributed call control using primitives such as those
      defined in REFER [RFC3515], Replaces [RFC3891], and Join

   o  Be mixing model independent.  The bulk of interesting multi-party
      applications involve mixing or combining media from multiple
      participants.  This mixing can be performed by one or more of the
      participants or by a centralized mixing resource.  The experience
      of the participants should not depend on the mixing model used.
      While most examples in this document refer to audio mixing, the
      framework applies to any media type.  In this context, a "mixer"
      refers to combining media of the same type in an appropriate,
      media-specific way.  This is consistent with the model described
      in the SIP conferencing framework.

   o  Be invoker oriented.  Only the user who invokes a feature or a
      service needs to know exactly which service is invoked or why.
      This is good because it allows new services to be created without
      requiring new primitives from all of the participants; and it
      allows for much simpler feature authorization policies, for
      example, when participation spans organizational boundaries.  As
      discussed in Section 2.7, this also avoids exponential state
      explosion when combining features.  The invoker only has to manage
      a user interface or application programming interface (API) to
      prevent local feature interactions.  All the other participants
      simply need to manage the feature interactions of a much smaller
      number of primitives.

   o  Primitives make full use of URIs (uniform resource identifiers).
      URIs are a very powerful mechanism for describing users and
      services.  They represent a plentiful resource that can be
      extremely expressive and easily routed, translated, and
      manipulated -- even across organizational boundaries.  URIs can
      contain special parameters and informational header fields that
      need only be relevant to the owner of the namespace (domain) of
      the URI.  Just as a user who selects an http: URL need not
      understand the significance and organization of the web site it
      references, a user may encounter a SIP URI that translates into an
      email-style group alias, which plays a pre-recorded message or
      runs some complex call-handling logic.  Note that while this may

Mahy, et al.                  Informational                     [Page 5]

RFC 5850               SIP Call Control Framework               May 2010

      seem paradoxical to the previous goal, both goals can be satisfied
      by the same model.

   o  Make use of SIP header fields and SIP event packages to provide
      SIP entities with information about their environment.  These
      should include information about the status/handling of dialogs on
      other user agents (UAs), information about the history of other
      contacts attempted prior to the current contact, the status of
      participants, the status of conferences, user presence
      information, and the status of messages.

   o  Encourage service decomposition, and design to make use of
      standard components using well-defined, simple interfaces.  Sample
      components include a SIP mixer, recording service, announcement
      server, and voice-dialog server.  (This is not an exhaustive

   o  Include authentication, authorization, policy, logging, and
      accounting mechanisms to allow these primitives to be used safely
      among mutually untrusted participants.  Some of these mechanisms
      may be used to assist in billing, but no specific billing system
      will be endorsed.

   o  Permit graceful fallback to baseline SIP.  Definitions for new SIP
      call control extensions/primitives must describe a graceful way to
      fallback to baseline SIP behavior.  Support for one primitive must
      not imply support for another primitive.

   o  Don't reinvent traditional models, such as the model used for the
      H.450 family of protocols, JTAPI (Java Telephony Application
      Programming Interface), or the CSTA (Computer-supported
      telecommunications applications) call model, as these other models
      do not share the design goals presented in this document.

   Note that the flexibility in this model does have some disadvantages
   in terms of interoperability.  It is possible to build a call control
   feature in SIP using different combinations of primitives.  For a
   discussion of the issues associated with this, see [BLISS-PROBLEM].

2.  Key Concepts

   This section introduces a number of key concepts that will be used to
   describe and explain various call control operations and services in
   the remainder of this document.  This includes the conversation space
   model, signaling and mixing models, common components, and the use of

Mahy, et al.                  Informational                     [Page 6]

RFC 5850               SIP Call Control Framework               May 2010

2.1.  Conversation Space Model

   This document introduces the concept of an abstract "conversation
   space" as a set of participants who believe they are all
   communicating among one another.  Each conversation space contains
   one or more participants.

   Participants are SIP UAs that send original media to or terminate and
   receive media from other members of the conversation space.
   Logically, every participant in the conversation space has access to
   all the media generated in that space (this is strictly true if all
   participants share a common media type).  A SIP UA that does not
   contribute or consume any media is NOT a participant, nor is a UA
   that merely forwards, transcodes, mixes, or selects media originating
   elsewhere in the conversation space.

      Note that a conversation space consists of zero or more SIP calls
      or SIP conferences.  A conversation space is similar to the
      definition of a "call" in some other call models.

   Participants may represent human users or non-human users (referred
   to as robots or automatons in this document).  Some participants may
   be hidden within a conversation space.  Some examples of hidden
   participants include: robots that generate tones, images, or
   announcements during a conference to announce users arriving and
   departing, a human call center supervisor monitoring a conversation
   between a trainee and a customer, and robots that record media for
   training or archival purposes.

   Participants may also be active or passive.  Active participants are
   expected to be intelligent enough to leave a conversation space when
   they no longer desire to participate.  (An attentive human
   participant is obviously active.)  Some robotic participants (such as
   a voice-messaging system, an instant-messaging agent, or a voice-
   dialog system) may be active participants if they can leave the
   conversation space when there is no human interaction.  Other robots
   (for example, our tone-generating robot from the previous example)
   are passive participants.  A human participant "on hold" is passive.

   An example diagram of a conversation space can be shown as a "bubble"
   or ovals, or as a "set" in curly or square bracket notation.  Each
   set, oval, or bubble represents a conversation space.  Hidden
   participants are shown in lowercase letters.  Examples are given in
   Figure 1.

   Note that while the term "conversation" usually applies to oral
   exchange of information, we apply the conversation space model to any
   media exchange between participants.

Mahy, et al.                  Informational                     [Page 7]

RFC 5850               SIP Call Control Framework               May 2010

   { A , B }                   [ A , b, C, D ]

      .-.                 .---.
     /   \               /     \
    /  A  \             / A   b \
   (       )           (         )
    \  B  /             \ C   D /
     \   /               \     /
      '-'                 '---'

   Figure 1.  Conversation Spaces

2.2.  Relationship between Conversation Space, SIP Dialogs, and SIP

   In [RFC3261], a call is "an informal term that refers to some
   communication between peers, generally set up for the purposes of a
   multimedia conversation".  The concept of a conversation space is
   needed because the SIP definition of call is not sufficiently precise
   for the purpose of describing the user experience of multi-party

   Do any other definitions convey the correct meaning?  SIP and SDP
   (Session Description Protocol) [RFC4566] both define a conference as
   "a multimedia session identified by a common session description".  A
   session is defined as "a set of multimedia senders and receivers and
   the data streams flowing from senders to receivers".  The definition
   of "call" in some call models is more similar to our definition of a
   conversation space.

   Some examples of the relationship between conversation spaces, SIP
   dialogs, and SIP sessions are listed below.  In each example, a human
   user will perceive that there is a single call.

   o  A simple two-party call is a single conversation space, a single
      session, and a single dialog.

   o  A locally mixed three-way call is two sessions and two dialogs.
      It is also a single conversation space.

   o  A simple dial-in audio conference is a single conversation space,
      but is represented by as many dialogs and sessions as there are
      human participants.

   o  A multicast conference is a single conversation space, a single
      session, and as many dialogs as participants.

Mahy, et al.                  Informational                     [Page 8]

RFC 5850               SIP Call Control Framework               May 2010

2.3.  Signaling Models

   Obviously, to make changes to a conversation space, you must be able
   to use SIP signaling to cause these changes.  Specifically, there
   must be a way to manipulate SIP dialogs (call legs) to move
   participants into and out of conversation spaces.  Although this is
   not as obvious, there also must be a way to manipulate SIP dialogs to
   include non-participant UAs that are otherwise involved in a
   conversation space (e.g., back-to-back user agents or B2BUAs, third
   party call control (3pcc) controllers, mixers, transcoders,
   translators, or relays).

   Implementations may setup the media relationships described in the
   conversation space model using a centralized control model.  One
   common way to implement this using SIP is known as third party call
   control (3pcc) and is described in 3pcc [RFC3725].  The 3pcc approach
   relies on only the following three primitive operations:

   o  Create a new dialog (INVITE)

   o  Modify a dialog (reINVITE)

   o  Destroy a dialog (BYE)

   The main advantage of the 3pcc approach is that it only requires very
   basic SIP support from end systems to support call control features.
   As such, third party call control is a natural way to handle protocol
   conversion and mid-call features.  It also has the advantage and
   disadvantage that new features can/must be implemented in one place
   only (the controller), and it neither requires enhanced client
   functionality nor takes advantage of it.

   In addition, a peer-to-peer approach is discussed at length in this
   document.  The primary drawback of the peer-to-peer model is
   additional complexity in the end system and authentication and
   management models.  The benefits of the peer-to-peer model include:

   o  state remains at the edges,

   o  call signaling need only go through participants involved (there
      are no additional points of failure), and

   o  peers may take advantage of end-to-end message integrity or

Mahy, et al.                  Informational                     [Page 9]

RFC 5850               SIP Call Control Framework               May 2010

   The peer-to-peer approach relies on additional "primitive"
   operations, some of which are identified here.

   o  Replace an existing dialog

   o  Join a new dialog with an existing dialog

   o  Locally perform media forking (multi-unicast)

   o  Ask another user agent (UA) to send a request on your behalf

   The peer-to-peer approach also only results in a single SIP dialog,
   directly between the two UAs.  The 3pcc approach results in two SIP
   dialogs, between each UA and the controller.  As a result, the SIP
   features and extensions that will be used during the dialog are
   limited to the those understood by the controller.  As a result, in a
   situation where both the UAs support an advanced SIP feature but the
   controller does not, the feature will not be able to be used.

   Many of the features, primitives, and actions described in this
   document also require some type of media mixing, combining, or
   selection as described in the next section.

2.4.  Mixing Models

   SIP permits a variety of mixing models, which are discussed here
   briefly.  This topic is discussed more thoroughly in the SIP
   conferencing framework [RFC4353] and [RFC4579].  SIP supports both
   tightly coupled and loosely coupled conferencing, although more
   sophisticated behavior is available in tightly coupled conferences.
   In a tightly coupled conference, a single SIP user agent (called the
   focus) has a direct dialog relationship with each participant (and
   may control non-participant user agents as well).  The focus can
   authoritatively publish information about the character and
   participants in a conference.  In a loosely coupled conference, there
   are no coordinated signaling relationships among the participants.

   For brevity, only the two most popular conferencing models are
   significantly discussed in this document (local and centralized
   mixing).  Applications of the conversation spaces model to loosely
   coupled multicast and distributed full unicast mesh conferences are
   left as an exercise for the reader.  Note that a distributed full
   mesh conference can be used for basic conferences, but does not
   easily allow for more complex conferencing actions like splitting,
   merging, and sidebars.

Mahy, et al.                  Informational                    [Page 10]

RFC 5850               SIP Call Control Framework               May 2010

   Call control features should be designed to allow a mixer (local or
   centralized) to decide when to reduce a conference back to a two-
   party call, or drop all the participants (for example, if only two
   automatons are communicating).  The actual heuristics used to release
   calls are beyond the scope of this document, but may depend on
   properties in the conversation space, such as the number of active,
   passive, or hidden participants and the send-only, receive-only, or
   send-and-receive orientation of various participants.

2.4.1.  Tightly Coupled

   Tightly coupled conferences utilize a central point for signaling and
   authentication known as a focus [RFC4353].  The actual media can be
   centrally mixed or distributed.  (Single) End System Mixing

   The first model we call "end system mixing".  In this model, user A
   calls user B, and they have a conversation.  At some point later, A
   decides to conference in user C.  To do this, A calls C, using a
   completely separate SIP call.  This call uses a different Call-ID,
   different tags, etc.  There is no call set up directly between B and
   C.  No SIP extension or external signaling is needed.  A merely
   decides to locally join two dialogs.

      B     C
       \   /
        \ /

   Figure 2.  End System Mixing Example

   In Figure 2, A receives media streams from both B and C, and mixes
   them.  A sends a stream containing A's and C's streams to B, and a
   stream containing A's and B's streams to C.  Basically, user A
   handles both signaling and media mixing.  Centralized Mixing

   In a centralized mixing model, all participants have a pairwise SIP
   and media relationship with the mixer.  Common applications of
   centralized mixing include ad hoc conferences and scheduled dial-in
   or dial-out conferences.  In Figure 3 below, the mixer M receives and
   sends media to participants A, B, C, D, and E.

Mahy, et al.                  Informational                    [Page 11]

RFC 5850               SIP Call Control Framework               May 2010

      B     C
       \   /
        \ /
         M --- A
        / \
       /   \
      D     E

   Figure 3.  Centralized Mixing Example  Centralized Signaling, Distributed Media

   In this conferencing model, there is a centralized controller, as in
   the dial-in and dial-out cases.  However, the centralized server
   handles signaling only.  The media is still sent directly between
   participants, using either multicast or multi-unicast.  Participants
   perform their own mixing.  Multi-unicast is when a user sends
   multiple packets (one for each recipient, addressed to that
   recipient).  This is referred to as a "Decentralized Multipoint
   Conference" in [H.323].  Full mesh media with centralized mixing is
   another approach.

2.4.2.  Loosely Coupled

   In these models, there is no point of central control of SIP
   signaling.  As in the "Centralized Signaling, Distributed Media" case
   above, all endpoints send media to all other endpoints.
   Consequently, every endpoint mixes their own media from all the other
   sources and sends their own media to every other participant.  Large-Scale Multicast Conferences

   Large-scale multicast conferences were the original motivation for
   both the Session Description Protocol (SDP) [RFC4566] and SIP.  In a
   large-scale multicast conference, one or more multicast addresses are
   allocated to the conference.  Each participant joins those multicast
   groups and sends their media to those groups.  Signaling is not sent
   to the multicast groups.  The sole purpose of the signaling is to
   inform participants of which multicast groups to join.  Large-scale
   multicast conferences are usually pre-arranged, with specific start
   and stop times.  However, multicast conferences do not need to be
   pre-arranged, so long as a mechanism exists to dynamically obtain a
   multicast address.

Mahy, et al.                  Informational                    [Page 12]

RFC 5850               SIP Call Control Framework               May 2010  Full Distributed Unicast Conferencing

   In this conferencing model, each participant has both a pairwise
   media relationship and a pairwise signaling relationship with every
   other participant (a full mesh).  This model requires a mechanism to
   maintain a consistent view of distributed state across the group.
   This is a classic, hard problem in computer science.  Also, this
   model does not scale well for large numbers of participants.  For <n>
   participants, the number of media and signaling relationships is
   approximately n-squared.  As a result, this model is not generally
   available in commercial implementations; to the contrary, it is
   primarily the topic of research or experimental implementations.
   Note that this model assumes peer-to-peer signaling.

2.5.  Conveying Information and Events

   Participants should have access to information about the other
   participants in a conversation space so that this information can be
   rendered to a human user or processed by an automaton.  Although some
   of this information may be available from the Request-URI or To,
   From, Contact, or other SIP header fields, another mechanism of
   reporting this information is necessary.

   Many applications are driven by knowledge about the progress of calls
   and conferences.  In general, these types of events allow for the
   construction of distributed applications, where the application
   requires information on dialog and conference state, but is not
   necessarily a co-resident with an endpoint user agent or conference
   server.  For example, a focus involved in a conversation space may
   wish to provide URIs for conference status and/or conference/floor

   The SIP Events architecture [RFC3265] defines general mechanisms for
   subscription to and notification of events within SIP networks.  It
   introduces the notion of a package that is a specific "instantiation"
   of the events mechanism for a well-defined set of events.

   Event packages are needed to provide the status of a user's dialogs,
   the status of conferences and their participants, user-presence
   information, the status of registrations, and the status of a user's
   messages.  While this is not an exhaustive list, these are sufficient
   to enable the sample features described in this document.

   The conference event package [RFC4575] allows users to subscribe to
   information about an entire tightly coupled SIP conference.
   Notifications convey information about the participants such as the
   SIP URI identifying each user, their status in the space (active,
   declined, departed), URIs to invoke other features (such as sidebar

Mahy, et al.                  Informational                    [Page 13]

RFC 5850               SIP Call Control Framework               May 2010

   conversations), links to other relevant information (such as floor-
   control policies), and if floor-control policies are in place, the
   user's floor-control status.  For conversation spaces created from
   cascaded conferences, conversation state can be gathered from
   relevant foci and merged into a cohesive set of state.

   The dialog package [RFC4235] provides information about all the
   dialogs the target user is maintaining, in which conversations the
   user is participating, and how these are correlated.  Likewise, the
   registration package [RFC3680] provides notifications when contacts
   have changed for a specific address-of-record (AOR).  The combination
   of these allows a user agent to learn about all conversations
   occurring for the entire registered contact set for an address-of-

   Note that user presence in SIP [RFC3856] has a close relationship
   with these latter two event packages.  It is fundamental to the
   presence model that the information used to obtain user presence is
   constructed from any number of different input sources.  Examples of
   other such sources include calendaring information and uploads of
   presence documents.  These two packages can be considered another
   mechanism that allows a presence agent to determine the presence
   state of the user.  Specifically, a user presence server can act as a
   subscriber for the dialog and registration packages to obtain
   additional information that can be used to construct a presence

   The multi-party architecture may also need to provide a mechanism to
   get information about the status/handling of a dialog (for example,
   information about the history of other contacts attempted prior to
   the current contact).  Finally, the architecture should provide ample
   opportunities to present informational URIs that relate to calls,
   conversations, or dialogs in some way.  For example, consider the SIP
   Call-Info header or Contact header fields returned in a 300-class
   response.  Frequently, additional information about a call or dialog
   can be fetched via non-SIP URIs.  For example, consider a web page
   for package tracking when calling a delivery company or a web page
   with related documentation when joining a dial-in conference.  The
   use of URIs in the multi-party framework is discussed in more detail
   in Section 3.7.

   Finally, the interaction of SIP with stimulus-signaling-based
   applications, which allow a user agent to interact with an
   application without knowledge of the semantics of that application,
   is discussed in the SIP application interaction framework [RFC5629].
   Stimulus signaling can occur with a user interface running locally
   with the client, or with a remote user interface, through media
   streams.  Stimulus signaling encompasses a wide range of mechanisms,

Mahy, et al.                  Informational                    [Page 14]

RFC 5850               SIP Call Control Framework               May 2010

   from clicking on hyperlinks, to pressing buttons, to traditional
   Dual-Tone Multi Frequency (DTMF) input.  In all cases, stimulus
   signaling is supported through the use of markup languages, which
   play a key role in that framework.

2.6.  Componentization and Decomposition

   This framework proposes a decomposed component architecture with a
   very loose coupling of services and components.  This means that a
   service (such as a conferencing server or an auto-attendant) need not
   be implemented as an actual server.  Rather, these services can be
   built by combining a few basic components in straightforward or
   arbitrarily complex ways.

   Since the components are easily deployed on separate boxes, by
   separate vendors, or even with separate providers, we achieve a
   separation of function that allows each piece to be developed in
   complete isolation.  We can also reuse existing components for new
   applications.  This allows rapid service creation, and the ability
   for services to be distributed across organizational domains anywhere
   in the Internet.

   For many of these components, it is also desirable to discover their
   capabilities, for example, querying the ability of a mixer to host a
   10-dialog conference or to reserve resources for a specific time.
   These actions could be provided in the form of URIs, provided there
   is an a priori means of understanding their semantics.  For example,
   if there is a published dictionary of operations, a way to query the
   service for the available operations and the associated URIs, the URI
   can be the interface for providing these service operations.  This
   concept is described in more detail in the context of dialog
   operations in Section 3.

2.6.1.  Media Intermediaries

   Media intermediaries are not participants in any conversation space,
   although an entity that is also a media translator may also have a
   co-located participant component (for example, a mixer that also
   announces the arrival of a new participant; the announcement portion
   is a participant, but the mixer itself is not).  Media intermediaries
   should be as transparent as possible to the end users -- offering a
   useful, fundamental service without getting in the way of new
   features implemented by participants.  Some common media
   intermediaries are described below.

Mahy, et al.                  Informational                    [Page 15]

RFC 5850               SIP Call Control Framework               May 2010  Mixer

   A SIP mixer is a component that combines media from all dialogs in
   the same conversation in a media-specific way.  For example, the
   default combining for an audio conference might be an N-1
   configuration, while a text mixer might interleave text messages on a
   per-line basis.  More details about how to manipulate the media
   policy used by mixers is discussed in [XCON-CCMP].  Transcoder

   A transcoder translates media from one encoding or format to another
   (for example, GSM (Global System for Mobile communications) voice to
   G.711, MPEG2 to H.261, or text/html to text/plain), or from one media
   type to another (for example, text to speech).  A more thorough
   discussion of transcoding is described in the SIP transcoding
   services invocation [RFC5369].  Media Relay

   A media relay terminates media and simply forwards it to a new
   destination without changing the content in any way.  Sometimes,
   media relays are used to provide source IP address anonymity, to
   facilitate middlebox traversal, or to provide a trusted entity where
   media can be forcefully disconnected.  Queue Server

   A queue server is a location where calls can be entered into one of
   several FIFO (first-in, first-out) queues.  A queue server would
   subscribe to the presence of groups or individuals who are interested
   in its queues.  When detecting that a user is available to service a
   queue, the server redirects or transfers the last call in the
   relevant queue to the available user.  On a queue-by-queue basis,
   authorized users could also subscribe to the call state (dialog
   information) of calls within a queue.  Authorized users could use
   this information to effectively pluck (take) a call out of the queue
   (for example, by sending an INVITE with a Replaces header to one of
   the user agents in the queue).  Parking Place

   A parking place is a location where calls can be terminated
   temporarily and then retrieved later.  While a call is "parked", it
   can receive media "on hold" such as music, announcements, or
   advertisements.  Such a service could be further decomposed such that
   announcements or music are handled by a separate component.

Mahy, et al.                  Informational                    [Page 16]

RFC 5850               SIP Call Control Framework               May 2010  Announcements and Voice Dialogs

   An announcement server is a server that can play digitized media
   (frequently audio), such as music or recorded speech.  These servers
   are typically accessible via SIP, HTTP (Hyper Text Transport
   Protocol), or RTSP (Real-Time Streaming Protocol).  An analogous
   service is a recording service that stores digitized media.  A
   convention for specifying announcements in SIP URIs is described in
   [RFC4240].  Likewise, the same server could easily provide a service
   that records digitized media.

   A "voice dialog" is a model of spoken interactive behavior between a
   human and an automaton that can include synthesized speech, digitized
   audio, recognition of spoken and DTMF key input, a recording of
   spoken input, and interaction with call control.  Voice dialogs
   frequently consist of forms or menus.  Forms present information and
   gather input; menus offer choices of what to do next.

   Spoken dialogs are a basic building block of applications that use
   voice.  Consider, for example, that a voicemail system, the
   conference-id and passcode collection system for a conferencing
   system, and complicated voice-portal applications all require a
   voice-dialog component.

2.6.2.  Text-to-Speech and Automatic Speech Recognition

   Text-to-speech (TTS) is a service that converts text into digitized
   audio.  TTS is frequently integrated into other applications, but
   when separated as a component, it provides greater opportunity for
   broad reuse.  Automatic Speech Recognition (ASR) is a service that
   attempts to decipher digitized speech based on a proposed grammar.
   Like TTS, ASR services can be embedded, or exposed so that many
   applications can take advantage of such services.  A standardized
   (decomposed) interface to access standalone TTS and ASR services is
   currently being developed as described in [RFC4313].

2.6.3.  VoiceXML

   VoiceXML is a W3C (World Wide Web Consortium) recommendation that was
   designed to give authors control over the spoken dialog between users
   and applications.  The application and user take turns speaking: the
   application prompts the user, and the user in turn responds.  Its
   major goal is to bring the advantages of web-based development and
   content delivery to interactive voice-response applications.  We
   believe that VoiceXML represents the ideal partner for SIP in the
   development of distributed IVR (interactive voice response) servers.
   VoiceXML is an XML-based scripting language for describing IVR
   services at an abstract level.  VoiceXML supports DTMF recognition,

Mahy, et al.                  Informational                    [Page 17]

RFC 5850               SIP Call Control Framework               May 2010

   speech recognition, text-to-speech, and the playing out of recorded
   media files.  The results of the data collected from the user are
   passed to a controlling entity through an HTTP POST operation.  The
   controller can then return another script, or terminate the
   interaction with the IVR server.

   A VoiceXML server also need not be implemented as a monolithic
   server.  Figure 4 shows a diagram of a VoiceXML browser that is split
   into media and non-media handling parts.  The VoiceXML interpreter
   handles SIP dialog state and state within a VoiceXML document, and
   sends requests to the media component over another protocol.

                       |             |
                       | VoiceXML    |
                       | Interpreter |
                       | (signaling) |
                         ^          ^
                         |          |
                     SIP |          | RTSP
                         |          |
                         |          |
                         v          v
            +-------------+        +-------------+
            |             |        |             |
            |  SIP UA     |   RTP  | RTSP Server |
            |             |<------>|   (media)   |
            |             |        |             |
            +-------------+        +-------------+

   Figure 4.  Decomposed VoiceXML Server

2.7.  Use of URIs

   All naming in SIP uses URIs.  URIs in SIP are used in a plethora of
   contexts: the Request-URI; Contact, To, From, and *-Info header
   fields; application/uri bodies; and embedded in email, web pages,
   instant messages, and ENUM records.  The Request-URI identifies the
   user or service for which the call is destined.

   SIP URIs embedded in informational SIP header fields, SIP bodies, and
   non-SIP content can also specify methods, special parameters, header
   fields, and even bodies.  For example:


Mahy, et al.                  Informational                    [Page 18]

RFC 5850               SIP Call Control Framework               May 2010

   Throughout this document, we discuss call control primitive
   operations.  One of the biggest problems is defining how these
   operations may be invoked.  There are a number of ways to do this.
   One way is to define the primitives in the protocol itself such that
   SIP methods (for example, REFER) or SIP header fields (for example,
   Replaces) indicate a specific call control action.  Another way to
   invoke call control primitives is to define a specific Request-URI
   naming convention.  Either these conventions must be shared between
   the client (the invoker) and the server, or published by or on behalf
   of the server.  The former involves defining URI construction
   techniques (e.g., URI parameters and/or token conventions) as
   proposed in [RFC4240].  The latter technique usually involves
   discovering the URI via a SIP event package, a web page, a business
   card, or an instant message.  Yet, another means to acquire the URIs
   is to define a dictionary of primitives with well-defined semantics
   and provide a means to query the named primitives and corresponding
   URIs that may be invoked on the service or dialogs.

2.7.1.  Naming Users in SIP

   An address-of-record, or public SIP address, is a SIP (or Secure SIP
   (SIPS)) URI that points to a domain with a location service that can
   map the URI to set of Contact URIs where the user might be available.
   Typically, the Contact URIs are populated via registration.

   Address-of-Record               Contacts

   sip:bob@biloxi.example.com -> sip:bob@babylon.biloxi.example.com:5060

   Callee Capabilities [RFC3840] define a set of additional parameters
   to the Contact header field that define the characteristics of the
   user agent at the specified URI.  For example, there is a mobility
   parameter that indicates whether the UA is fixed or mobile.  When a
   user agent registers, it places these parameters in the Contact
   header fields to characterize the URIs it is registering.  This
   allows a proxy for that domain to have information about the contact
   addresses for that user.

   When a caller sends a request, it can optionally request Caller
   Preferences [RFC3841] by including the Accept-Contact, Request-
   Disposition, and Reject-Contact header fields that request certain
   handling by the proxy in the target domain.  These header fields
   contain preferences that describe the set of desired URIs to which
   the caller would like their request routed.  The proxy in the target
   domain matches these preferences with the Contact characteristics
   originally registered by the target user.  The target user can also

Mahy, et al.                  Informational                    [Page 19]

RFC 5850               SIP Call Control Framework               May 2010

   choose to run arbitrarily complex "Find-me" feature logic on a proxy
   in the target domain.

   There is a strong asymmetry in how preferences for callers and
   callees can be presented to the network.  While a caller takes an
   active role by initiating the request, the callee takes a passive
   role in waiting for requests.  This motivates the use of callee-
   supplied scripts and caller preferences included in the call request.
   This asymmetry is also reflected in the appropriate relationship
   between caller and callee preferences.  A server for a callee should
   respect the wishes of the caller to avoid certain locations, while
   the preferences among locations has to be the callee's choice, as it
   determines where, for example, the phone rings and whether the callee
   incurs mobile telephone charges for incoming calls.

   SIP User Agent implementations are encouraged to make intelligent
   decisions based on the type of participants (active/passive, hidden,
   human/robot) in a conversation space.  This information is conveyed
   via the dialog package or in a SIP header field parameter
   communicated using an appropriate SIP header field.  For example, a
   music on hold service may take the sensible approach that if there
   are two or more unhidden participants, it should not provide hold
   music; or that it will not send hold music to robots.

   Multiple participants in the same conversation space may represent
   the same human user.  For example, the user may use one participant
   device for video, chat, and whiteboard media on a PC and another for
   audio media on a SIP phone.  In this case, the address-of-record is
   the same for both user agents, but the Contacts are different.  In
   this case, there is really only one human participant.  In addition,
   human users may add robot participants that act on their behalf (for
   example, a call recording service or a calendar announcement
   reminder).  Call control features in SIP should continue to function
   as expected in such an environment.

2.7.2.  Naming Services with SIP URIs

   A critical piece of defining a session-level service that can be
   accessed by SIP is defining the naming of the resources within that
   service.  This point cannot be overstated.

   In the context of SIP control of application components, we take
   advantage of the fact that the left-hand side of a standard SIP URI
   is a user part.  Most services may be thought of as user automatons
   that participate in SIP sessions.  It naturally follows that the user
   part should be utilized as a service indicator.

Mahy, et al.                  Informational                    [Page 20]

RFC 5850               SIP Call Control Framework               May 2010

   For example, media servers commonly offer multiple services at a
   single host address.  Use of the user part as a service indicator
   enables service consumers to direct their requests without ambiguity.
   It has the added benefit of enabling media services to register their
   availability with SIP Registrars just as any "real" SIP user would.
   This maintains consistency and provides enhanced flexibility in the
   deployment of media services in the network.

   There has been much discussion about the potential for confusion if
   media-service URIs are not readily distinguishable from other types
   of SIP UAs.  The use of a service namespace provides a mechanism to
   unambiguously identify standard interfaces while not constraining the
   development of private or experimental services.

   In SIP, the Request-URI identifies the user or service for which the
   call is destined.  The great advantage of using URIs (specifically,
   the SIP Request-URI) as a service identifier comes because of the
   combination of two facts.  First, unlike in the PSTN (Public Switched
   Telephone Network), where the namespace (dialable telephone numbers)
   is limited, URIs come from an infinite space.  They are plentiful,
   and they are free.  Secondly, the primary function of SIP is call
   routing through manipulations of the Request-URI.  In the traditional
   SIP application, this URI represents a person.  However, the URI can
   also represent a service, as we propose here.  This means we can
   apply the routing services SIP provides to the routing of calls to
   services.  The result -- the problem of service invocation and
   service location becomes a routing problem, for which SIP provides a
   scalable and flexible solution.  Since there is such a vast namespace
   of services, we can explicitly name each service in a finely granular
   way.  This allows the distribution of services across the network.
   For further discussion about services and SIP URIs, see RFC 3087

   Consider a conferencing service, where we have separated the names of
   ad hoc conferences from scheduled conferences, we can program proxies
   to route calls for ad hoc conferences to one set of servers and calls
   for scheduled ones to another, possibly even in a different provider.
   In fact, since each conference itself is given a URI, we can
   distribute conferences across servers, and easily guarantee that
   calls for the same conference always get routed to the same server.
   This is in stark contrast to conferences in the telephone network,
   where the equivalent of the URI -- the phone number -- is scarce.  An
   entire conferencing provider generally has one or two numbers.
   Conference IDs must be obtained through IVR interactions with the
   caller or through a human attendant.  This makes it difficult to
   distribute conferences across servers all over the network, since the
   PSTN routing only knows about the dialed number.

Mahy, et al.                  Informational                    [Page 21]

RFC 5850               SIP Call Control Framework               May 2010

   For more examples, consider the URI conventions of RFC 4240 [RFC4240]
   for media servers and RFC 4458 [RFC4458] for voicemail and IVR

   In practical applications, it is important that an invoker does not
   necessarily apply semantic rules to various URIs it did not create.
   Instead, it should allow any arbitrary string to be provisioned, and
   map the string to the desired behavior.  The administrator of a
   service may choose to provision specific conventions or mnemonic
   strings, but the application should not require it.  In any large
   installation, the system owner is likely to have preexisting rules
   for mnemonic URIs, and any attempt by an application to define its
   own rules may create a conflict.  Implementations should allow an
   arbitrary mix of URIs from these schemes, or any other scheme that
   renders valid SIP URIs, rather than enforce only one particular

   As we have shown, SIP URIs represent an ideal, flexible mechanism for
   describing and naming service resources, regardless of whether the
   resources are queues, conferences, voice dialogs, announcements,
   voicemail treatments, or phone features.

2.8.  Invoker Independence

   With functional signaling, only the invoker of features in SIP needs
   to know exactly which feature they are invoking.  One of the primary
   benefits of this approach is that combinations of functional features
   work in SIP call control without requiring complex feature-
   interaction matrices.  For example, let us examine the combination of
   a "transfer" of a call that is "conferenced".

   Alice calls Bob.  Alice silently "conferences in" her robotic
   assistant Albert as a hidden party.  Bob transfers Alice to Carol.
   If Bob asks Alice to Replace her leg with a new one to Carol, then
   both Alice and Albert should be communicating with Carol

   Using the peer-to-peer model, this combination of features works fine
   if A is doing local mixing (Alice replaces Bob's dialog with
   Carol's), or if A is using a central mixer (the mixer replaces Bob's
   dialog with Carol's).  A clever implementation using the 3pcc model
   can generate similar results.

   New extensions to the SIP Call Control Framework should attempt to
   preserve this property.

Mahy, et al.                  Informational                    [Page 22]

RFC 5850               SIP Call Control Framework               May 2010

2.9.  Billing Issues

   Billing in the PSTN is typically based on who initiated a call.  At
   the moment, billing in a SIP network is neither consistent with
   itself nor with the PSTN.  (A billing model for SIP should allow for
   both PSTN-style billing and non-PSTN billing.)  The example below
   demonstrates one such inconsistency.

   Alice places a call to Bob.  Alice then blind transfers Bob to Carol
   through a PSTN gateway.  In current usage of REFER, Bob may be billed
   for a call he did not initiate (his UA originated the outgoing
   dialog, however).  This is not necessarily a terrible thing, but it
   demonstrates a security concern (Bob must have appropriate local
   policy to prevent fraud).  Also, Alice may wish to pay for Bob's
   session with Carol.  There should be a way to signal this in SIP.

   Likewise, a Replacement call may maintain the same billing
   relationship as a Replaced call, so if Alice first calls Carol, then
   asks Bob to Replace this call, Alice may continue to receive a bill.

   Further work in SIP billing should define a way to set or discover
   the direction of billing.

3.  Catalog of Call Control Actions and Sample Features

   Call control actions can be categorized by the dialogs upon which
   they operate.  The actions may involve a single or multiple dialogs.
   These dialogs can be early or established.  Multiple dialogs may be
   related in a conversation space to form a conference or other
   interesting media topologies.

   It should be noted that it is desirable to provide a means by which a
   party can discover the actions that may be performed on a dialog.
   The interested party may be independent or related to the dialogs.
   One means of accomplishing this is through the ability to define and
   obtain URIs for these actions, as described in Section 2.7.2.

   Below are listed several call control "actions" that establish or
   modify dialogs and relate the participants in a conversation space.
   The names of the actions listed are for descriptive purposes only
   (they are not normative).  This list of actions is not meant to be

   In the examples, all actions are initiated by the user "Alice"
   represented by UA "A".

Mahy, et al.                  Informational                    [Page 23]

RFC 5850               SIP Call Control Framework               May 2010

3.1.  Remote Call Control Actions on Early Dialogs

   The following are a set of actions that may be performed on a single
   early dialog.  These actions can be thought of as a set of remote
   control operations.  For example, an automaton might perform the
   operation on behalf of a user.  Alternatively, a user might use the
   remote control in the form of an application to perform the action on
   the early dialog of a UA that may be out of reach.  All of these
   actions correspond to telling the UA how to respond to a request to
   establish an early dialog.  These actions provide useful
   functionality for PDA-, PC-, and server-based applications that
   desire the ability to control a UA.  A proposed mechanism for this
   type of functionality is described in remote call control

3.1.1.  Remote Answer

   A dialog is in some early dialog state such as 180 Ringing.  It may
   be desirable to tell the UA to answer the dialog.  That is, tell it
   to send a 200 OK response to establish the dialog.

3.1.2.  Remote Forward or Put

   It may be desirable to tell the UA to respond with a 3xx class
   response to forward an early dialog to another UA.

3.1.3.  Remote Busy or Error Out

   It may be desirable to instruct the UA to send an error response such
   as 486 Busy Here.

3.2.  Remote Call Control Actions on Single Dialogs

   There is another useful set of actions that operate on a single
   established dialog.  These operations are useful in building
   productivity applications for aiding users in controlling their
   phones.  For example, a Customer Relationship Management (CRM)
   application that sets up calls for a user eliminating the need for
   the user to actually enter an address.  These operations can also be
   thought of as remote control actions.  A proposed mechanism for this
   type of functionality is described in remote call control

3.2.1.  Remote Dial

   This action instructs the UA to initiate a dialog.  This action can
   be performed using the REFER method.

Mahy, et al.                  Informational                    [Page 24]

RFC 5850               SIP Call Control Framework               May 2010

3.2.2.  Remote On and Off Hold

   This action instructs the UA to put an established dialog on hold.
   Though this operation can conceptually be performed with the REFER
   method, there are no semantics defined as to what the referred party
   should do with the SDP.  There is no way to distinguish between the
   desire to go on or off hold on a per-media stream basis.

3.2.3.  Remote Hangup

   This action instructs the UA to terminate an early or established
   dialog.  A REFER request with the following Refer-To URI and Target-
   Dialog header field [RFC4538] performs this action.  Note: this
   example does not show the full set of header fields.

   REFER sip:carol@client.chicago.net SIP/2.0
   Refer-To: sip:bob@babylon.biloxi.example.com;method=BYE
   Target-Dialog: 13413098;local-tag=879738;remote-tag=023214

3.3.  Call Control Actions on Multiple Dialogs

   These actions apply to a set of related dialogs.

3.3.1.  Transfer

   This section describes how call transfer can be achieved using
   centralized (3pcc) and peer-to-peer (REFER) approaches.

   The conversation space changes as follows:

    before            after
   { A , B }  -->   { C , B }

   A replaces itself with C.

   To make this happen using the peer-to-peer approach, "A" would send
   two SIP requests.  A shorthand for those requests is shown below:

   REFER B  Refer-To:C
   BYE B

   To make this happen using the 3pcc approach instead, the controller
   sends requests represented by the shorthand below:

   INVITE C (w/SDP of B)
   reINVITE B (w/SDP of C)
   BYE A

Mahy, et al.                  Informational                    [Page 25]

RFC 5850               SIP Call Control Framework               May 2010

   Features enabled by this action:

   - blind transfer
   - transfer to a central mixer (some type of conference or forking)
   - transfer to park server (park)
   - transfer to music on hold or announcement server
   - transfer to a "queue"
   - transfer to a service (such as voice-dialog service)
   - transition from local mixer to central mixer

   This action is frequently referred to as "completing an attended
   transfer".  It is described in more detail in [RFC5589].

   Note that if a transfer requires URI hiding or privacy, then the 3pcc
   approach can more easily implement this.  For example, if the URI of
   C needs to be hidden from B, then the use of 3pcc helps accomplish

3.3.2.  Take

   The conversation space changes as follows:

   { B , C } --> { B , A }

   A forcibly replaces C with itself.  In most uses of this primitive, A
   is just "un-replacing" itself.

   Using the peer-to-peer approach, "A" sends:

    INVITE B  Replaces: <dialog between B and C>

   Using the 3pcc approach (all requests sent from controller):

    INVITE A (w/SDP of B)
    reINVITE B (w/SDP of A)
    BYE C

   Features enabled by this action:

   - transferee completes an attended transfer
   - retrieve from central mixer (not recommended)
   - retrieve from music on hold or park
   - retrieve from queue
   - call center take
   - voice portal resuming ownership of a call it originated
   - answering-machine style screening (pickup)
   - pickup of a ringing call (i.e., early dialog)

Mahy, et al.                  Informational                    [Page 26]

RFC 5850               SIP Call Control Framework               May 2010

   Note that pick up of a ringing call has perhaps some interesting
   additional requirements.  First of all, it is an early dialog as
   opposed to an established dialog.  Secondly, the party that is to
   pick up the call may only wish to do so only while it is an early
   dialog.  That is in the race condition where the ringing UA accepts
   just before it receives signaling from the party wishing to take the
   call, the taking party wishes to yield or cancel the take.  The goal
   is to avoid yanking an answered call from the called party.

   This action is described in Replaces [RFC3891] and in [RFC5589].

3.3.3.  Add

   Note that the following four actions are described in [RFC4579].

   This is merely adding a participant to a SIP conference.  The
   conversation space changes as follows:

   { A , B } --> { A , B , C }

   A adds C to the conversation.

   Using the peer-to-peer approach, adding a party using local mixing
   requires no signaling.  To transition from a two-party call or a
   locally mixed conference to central mixing, A could send the
   following requests:

    REFER B  Refer-To: conference-URI
    INVITE conference-URI
    BYE B

   To add a party to a conference:

    REFER C  Refer-To: conference-URI
    REFER conference-URI  Refer-To: C

   Using the 3pcc approach to transition to centrally mixed, the
   controller would send:

    INVITE mixer leg 1 (w/SDP of A)
    INVITE mixer leg 2 (w/SDP of B)
    INVITE C (late SDP)
    reINVITE A (w/SDP of mixer leg 1)
    reINVITE B (w/SDP of mixer leg 2)
    INVITE mixer leg3 (w/SDP of C)

Mahy, et al.                  Informational                    [Page 27]

RFC 5850               SIP Call Control Framework               May 2010

   To add a party to a SIP conference:

    INVITE C (late SDP)
    INVITE conference-URI (w/SDP of C)

   Features enabled:

   - standard conference feature
   - call recording
   - answering-machine style screening (screening)

3.3.4.  Local Join

   The conversation space changes like this:

   { A , B } , { A , C }  -->  { A , B , C }

           or like this

   { A , B } , { C , D }  -->  { A , B , C , D }

   A takes two conversation spaces and joins them together into a single

   Using the peer-to-peer approach, A can mix locally, or REFER the
   participants of both conversation spaces to the same central mixer
   (as in Section 3.3.5).

   For the 3pcc approach, the call flows for inserting participants, and
   joining and splitting conversation spaces are tedious yet
   straightforward, so these are left as an exercise for the reader.

   Features enabled:

   - standard conference feature
   - leaving a sidebar to rejoin a larger conference

3.3.5.  Insert

   The conversation space changes like this:

   { B , C } --> { A , B , C }

   A inserts itself into a conversation space.

   A proposed mechanism for signaling this using the peer-to-peer
   approach is to send a new header field in an INVITE with "joining"
   [RFC3911] semantics.  For example:

Mahy, et al.                  Informational                    [Page 28]

RFC 5850               SIP Call Control Framework               May 2010

   INVITE B Join: <dialog id of B and C>

   If B accepted the INVITE, B would accept responsibility to set up the
   dialogs and mixing necessary (for example, to mix locally or to
   transfer the participants to a central mixer).

   Features enabled:

   - barge-in
   - call center monitoring
   - call recording

3.3.6.  Split

   { A , B , C , D } --> { A , B } , { C , D }

   If using a central conference with peer-to-peer

    REFER C  Refer-To: conference-URI (new URI)
    REFER D  Refer-To: conference-URI (new URI)
    BYE C
    BYE D

   Features enabled:

   - sidebar conversations during a larger conference

3.3.7.  Near-Fork

   A participates in two conversation spaces simultaneously:

   { A, B } --> { B , A } & { A , C }

   A is a participant in two conversation spaces such that A sends the
   same media to both spaces, and renders media from both spaces,
   presumably by mixing or rendering the media from both.  We can define
   that A is the "anchor" point for both forks, each of which is a
   separate conversation space.

   This action is purely local implementation (it requires no special
   signaling).  Local features such as switching calls between the
   background and foreground are possible using this media relationship.

3.3.8.  Far-Fork

   The conversation space diagram.

   { A, B } --> { A , B } & { B , C }

Mahy, et al.                  Informational                    [Page 29]

RFC 5850               SIP Call Control Framework               May 2010

   A requests B to be the "anchor" of two conversation spaces.

   This is easily set up by creating a conference with two sub-
   conferences and setting the media policy appropriately such that B is
   a participant in both.  Media forking can also be set up using 3pcc,
   as described in Section 5.1 of RFC 3264 [RFC3264] (an offer/answer
   model for SDP).  The session descriptions for forking are quite
   complex.  Controllers should verify that endpoints can handle forked
   media, for example, using prior configuration.

   Features enabled:

   - barge-in
   - voice-portal services
   - whisper
   - key word detection
   - sending DTMF somewhere else

4.  Security Considerations

   Call control primitives provide a powerful set of features that can
   be dangerous in the hands of an attacker.  To complicate matters,
   call control primitives are likely to be automatically authorized
   without direct human oversight.

   The class of attacks that are possible using these tools includes the
   ability to eavesdrop on calls, disconnect calls, redirect calls,
   render irritating content (including ringing) at a user agent, cause
   an action that has billing consequences, subvert billing (theft-of-
   service), and obtain private information.  Call control extensions
   must take extra care to describe how these attacks will be prevented.

   We can also make some general observations about authorization and
   trust with respect to call control.  The security model is
   dramatically dependent on the signaling model chosen (see Section

   Let us first examine the security model used in the 3pcc approach.
   All signaling goes through the controller, which is a trusted entity.
   Traditional SIP authentication and hop-by-hop encryption and message
   integrity work fine in this environment, but end-to-end encryption
   and message integrity may not be possible.

   When using the peer-to-peer approach, call control actions and
   primitives can be legitimately initiated by a) an existing
   participant in the conversation space, b) a former participant in the
   conversation space, or c) an entity trusted by one of the
   participants.  For example, a participant always initiates a

Mahy, et al.                  Informational                    [Page 30]

RFC 5850               SIP Call Control Framework               May 2010

   transfer; a retrieve from park (a take) is initiated on behalf of a
   former participant, and a barge-in (insert or far-fork) is initiated
   by a trusted entity (an operator, for example).

   Authenticating requests by an existing participant or a trusted
   entity can be done with baseline SIP mechanisms.  In the case of
   features initiated by a former participant, these should be protected
   against replay attacks, e.g., by using a unique name or identifier
   per invocation.  The Replaces header field exhibits this behavior as
   a by-product of its operation (once a Replaces operation is
   successful, the dialog being Replaced no longer exists).  These
   credentials may, for example, need to be passed transitively or
   fetched in an event body.

   To authorize call control primitives that trigger special behavior
   (such as an INVITE with Replaces or Join semantics), the receiving
   user agent may have trouble finding appropriate credentials with
   which to challenge or authorize the request, as the sender may be
   completely unknown to the receiver, except through the introduction
   of a third party.  These credentials need to be passed transitively
   in some way or fetched in an event body, for example.

   Standard SIP privacy and anonymity mechanisms such as [RFC3323] and
   [RFC3325] used during SIP session establishment apply equally well to
   SIP call control operations.  SIP call control mechanisms should
   address privacy and anonymity issues associated with that operation.
   For example, privacy during a transfer operation using REFER is
   discussed in Section 7.2 of [RFC5589]

Mahy, et al.                  Informational                    [Page 31]

RFC 5850               SIP Call Control Framework               May 2010

Appendix A.  Example Features

   Primitives are defined in terms of their ability to provide features.
   These example features should require an amply robust set of services
   to demonstrate a useful set of primitives.  They are described here
   briefly.  Note that the descriptions of these features are non-
   normative.  Note also that this document describes a mixture of both
   features originating in the world of telephones and features that are
   clearly Internet oriented.

Appendix A.1.  Attended Transfer

   In Attended Transfer [RFC5589], the transferring party establishes a
   session with the transfer target before completing the transfer.

Appendix A.2.  Auto Answer

   In Auto Answer, calls to a certain address or URI answer immediately
   via a speakerphone.  The Answer-Mode header field [RFC5373] can be
   used for this feature.

Appendix A.3.  Automatic Callback

   In Automatic Callback [RFC5359], Alice calls Bob, but Bob is busy.
   Alice would like Bob to call her automatically when he is available.
   When Bob hangs up, Alice's phone rings.  When Alice answers, Bob's
   phone rings.  Bob answers and they talk.

Appendix A.4.  Barge-In

   In Barge-in, Carol interrupts Alice who has an in-progress call with
   Bob.  In some variations, Alice forcibly joins a new conversation
   with Carol, in other variations, all three parties are placed in the
   same conversation (basically a three-way conference).  Barge-in works
   the same as call monitoring except that it must indicate that the
   send media stream be mixed so that all of the other parties can hear
   the stream from the UA that is barging in.

Appendix A.5.  Blind Transfer

   In Blind Transfer [RFC5589], Alice is in a conversation with Bob.
   Alice asks Bob to contact Carol, but makes no attempt to contact
   Carol independently.  In many implementations, Alice does not verify
   Bob's success or failure in contacting Carol.

Mahy, et al.                  Informational                    [Page 32]

RFC 5850               SIP Call Control Framework               May 2010

Appendix A.6.  Call Forwarding

   In call forwarding [RFC5359], before a dialog is accepted, it is
   redirected to another location, for example, because the originally
   intended recipient is busy, does not answer, is disconnected from the
   network, or has configured all requests to go elsewhere.

Appendix A.7.  Call Monitoring

   Call monitoring is a Join operation [RFC3911].  For example, a call
   center supervisor joins an in-progress call for monitoring purposes.
   The monitoring UA sends a Join to the dialog to which it wants to
   listen.  It is able to discover the dialog via the dialog state on
   the monitored UA.  The monitoring UA sends SDP in the INVITE that
   indicates receive-only media.  As the UA is only monitoring, it does
   not matter whether the UA indicates it wishes the send stream to be
   mixed or point to point.

Appendix A.8.  Call Park

   In Call Park [RFC5359], a participant parks a call (essentially puts
   the call on hold), and then retrieves it at a later time (typically
   from another location).  Call park requires the ability to put a
   dialog some place, advertise it to users in a pickup group, and to
   uniquely identify it in a means that can be communicated (including
   human voice).  The dialog can be held locally on the UA parking the
   dialog or alternatively transferred to the park service for the
   pickup group.  The parked dialog then needs to be labeled (e.g.,
   orbit 12) in a way that can be communicated to the party that is to
   pick up the call.  The UAs in the pickup group discover the parked
   dialog(s) via the dialog package from the park service.  If the
   dialog is parked locally, the park service merely aggregates the
   parked call states from the set of UAs in the pickup group.

Appendix A.9.  Call Pickup

   There are two different features that are called Call Pickup
   [RFC5359].  The first is the pickup of a parked dialog.  The UA from
   which the dialog is to be picked up subscribes to the dialog state of
   the park service or the UA that has locally parked the dialog.
   Dialogs that are parked should be labeled with an identifier.  The
   labels are used by the UA to allow the user to indicate which dialog
   is to be picked up.  The UA picking up the call invoked the URI in
   the call state that is labeled as replace-remote.

   The other call pickup feature involves picking up an early dialog
   (typically ringing).  A party picks up a call that was ringing at
   another location.  One variation allows the caller to choose which

Mahy, et al.                  Informational                    [Page 33]

RFC 5850               SIP Call Control Framework               May 2010

   location, another variation just picks up any call in that user's
   "pickup group".  This feature uses some of the same primitives as the
   pickup of a parked call.  The call state of the UA ringing phone is
   advertised using the dialog package.  The UA that is to pick up the
   early dialog subscribes either directly to the ringing UA or to a
   service aggregating the states for UAs in the pickup group.  The call
   state identifies early dialogs.  The UA uses the call state(s) to
   help the user choose which early dialog is to be picked up.  The UA
   then invokes the URI in the call state labeled as replace-remote.

Appendix A.10.  Call Return

   In Call Return, Alice calls Bob.  Bob misses the call or is
   disconnected before he is finished talking to Alice.  Bob invokes
   Call return, which calls Alice, even if Alice did not provide her
   real identity or location to Bob.

Appendix A.11.  Call Waiting

   In Call Waiting, Alice is in a call, then receives another call.
   Alice can place the first call on hold, and talk with the other
   caller.  She can typically switch back and forth between the callers.

Appendix A.12.  Click-to-Dial

   In Click-to-Dial [RFC5359], Alice looks in her company directory for
   Bob.  When she finds Bob, she clicks on a URI to call him.  Her phone
   rings (or possibly answers automatically), and when she answers,
   Bob's phone rings.  The application or server that hosts the Click-
   to-Dial application captures the URI to be dialed and can set up the
   call using 3pcc or can send a REFER request to the UA that is to dial
   the address.  As users sometimes change their mind or wish to give up
   listing to a ringing or voicemail answered phone, this application
   illustrates the need to also have the ability to remotely hangup a

Appendix A.13.  Conference Call

   In a Conference Call [RFC4579], there are three or more active,
   visible participants in the same conversation space.

Appendix A.14.  Consultative Transfer

   In Consultative Transfer [RFC5589], the transferring party
   establishes a session with the target and mixes both sessions
   together so that all three parties can participate, then disconnects
   leaving the transferee and transfer target with an active session.

Mahy, et al.                  Informational                    [Page 34]

RFC 5850               SIP Call Control Framework               May 2010

Appendix A.15.  Distinctive Ring

   In Distinctive Ring, incoming calls have different ring cadences or
   sample sounds depending on the From party, the To party, or other
   factors.  The target UA either makes a local decision based on
   information in an incoming INVITE (To, From, Contact, Request-URI) or
   trusts an Alert-Info header field [RFC3261] provided by the caller or
   inserted by a trusted proxy.  In the latter case, the UA fetches the
   content described in the URI (typically via HTTP) and renders it to
   the user.

Appendix A.16.  Do Not Disturb

   In Do Not Disturb, Alice selects the Do Not Disturb option.  Calls to
   her either ring briefly or not at all and are forwarded elsewhere.
   Some variations allow specially authorized callers to override this
   feature and ring Alice anyway.  Do Not Disturb is best implemented in
   SIP using presence [RFC3856].

Appendix A.17.  Find-Me

   In Find-Me, Alice sets up complicated rules for how she can be
   reached (possibly using CPL (Call Processing Language) [RFC3880],
   presence [RFC3856], or other factors).  When Bob calls Alice, his
   call is eventually routed to a temporary Contact where Alice happens
   to be available.

Appendix A.18.  Hotline

   In Hotline, Alice picks up a phone and is immediately connected to
   the technical support hotline, for example.  Hotline is also
   sometimes known as a Ringdown line.

Appendix A.19.  IM Conference Alerts

   In IM Conference Alerts, a user receives a notification as an instant
   message whenever someone joins a conference in which they are already
   a participant.

Appendix A.20.  Inbound Call Screening

   In Inbound Call Screening, Alice doesn't want to receive calls from
   Matt.  Inbound Screening prevents Matt from disturbing Alice.  In
   some variations, this works even if Matt hides his identity.

Mahy, et al.                  Informational                    [Page 35]

RFC 5850               SIP Call Control Framework               May 2010

Appendix A.21.  Intercom

   In Intercom, Alice typically presses a button on a phone that
   immediately connects to another user or phone and causes that phone
   to play her voice over its speaker.  Some variations immediately set
   up two-way communications, other variations require another button to
   be pressed to enable a two-way conversation.  The UA initiates a
   dialog using INVITE and the Answer-Mode: Auto header field as
   described in [RFC5373].  The called UA accepts the INVITE with a 200
   OK and automatically enables the speakerphone.

   Alternatively, this can be a local decision for the UA to auto answer
   based upon called-party identification.

Appendix A.22.  Message Waiting

   In Message Waiting [RFC3842], Bob calls Alice when she has stepped
   away from her phone.  When she returns, a visible or audible
   indicator conveys that someone has left her a voicemail message.  The
   message waiting indication may also convey how many messages are
   waiting, from whom, at what time, and other useful pieces of

Appendix A.23.  Music on Hold

   In Music on Hold [RFC5359], when Alice places a call with Bob on
   hold, it replaces its audio with streaming content such as music,
   announcements, or advertisements.  Music on hold can be implemented a
   number of ways.  One way is to transfer the held call to a holding
   service.  When the UA wishes to take the call off hold, it basically
   performs a take on the call from the holding service.  This involves
   subscribing to call state on the holding service and then invoking
   the URI in the call state labeled as replace-remote.

   Alternatively, music on hold can be performed as a local mixing
   operation.  The UA holding the call can mix in the music from the
   music service via RTP (i.e., an additional dialog) or RTSP or other
   streaming media source.  This approach is simpler (i.e., the held
   dialog does not move so there is less chance of loosing them) from a
   protocol perspective, however it does use more LAN bandwidth and
   resources on the UA.

Appendix A.24.  Outbound Call Screening

   In Outbound Call Screening, Alice is paged and unknowingly calls a
   PSTN pay-service telephone number in the Caribbean, but local policy
   blocks her call, and possibly informs her why.

Mahy, et al.                  Informational                    [Page 36]

RFC 5850               SIP Call Control Framework               May 2010

Appendix A.25.  Pre-Paid Calling

   In Pre-paid Calling, Alice pays for a certain currency or unit amount
   of calling value.  When she places a call, she provides her account
   number somehow.  If her account runs out of calling value during a
   call, her call is disconnected or redirected to a service where she
   can purchase more calling value.

   For prepaid calling, the user's media always passes through a device
   that is trusted by the pre-paid provider.  This may be the other
   endpoint (for example, a PSTN gateway).  In either case, an
   intermediary proxy or B2BUA can periodically verify the amount of
   time available on the pre-paid account, and use the session-timer
   extension to cause the trusted endpoint (gateway) or intermediary
   (media relay) to send a reINVITE before that time runs out.  During
   the reINVITE, the SIP intermediary can re-verify the account and
   insert another session-timer header field.

   Note that while most pre-paid systems on the PSTN use an IVR to
   collect the account number and destination, this isn't strictly
   necessary for a SIP-originated prepaid call.  SIP requests and SIP
   URIs are sufficiently expressive to convey the final destination, the
   provider of the prepaid service, the location from which the user is
   calling, and the prepaid account they want to use.  If a pre-paid IVR
   is used, the mechanism described below (Voice Portals) can be
   combined as well.

Appendix A.26.  Presence-Enabled Conferencing

   In Presence-Enabled Conferencing, Alice wants to set up a conference
   call with Bob and Cathy when they all happen to be available (rather
   than scheduling a predefined time).  The server providing the
   application monitors their status, and calls all three when they are
   all "online", not idle, and not in another call.  This could be
   implemented using conferencing [RFC4579] and presence [RFC3264]

Appendix A.27.  Single Line Extension/Multiple Line Appearance

   In Single Line Extension/Multiple Line Appearances, groups of phones
   are all treated as "extensions" of a single line or AOR.  A call for
   one rings them all.  As soon as one answers, the others stop ringing.
   If any extension is actively in a conversation, another extension can
   "pick up" and immediately join the conversation.  This emulates the
   behavior of a home telephone line with multiple phones.  Incoming
   calls ring all the extensions through basic parallel forking.  Each
   extension subscribes to dialog events from each other extension.
   While one user has an active call, any other UA extension can insert

Mahy, et al.                  Informational                    [Page 37]

RFC 5850               SIP Call Control Framework               May 2010

   itself into that conversation (it already knows the dialog
   information) in the same way as barge-in.

   When implemented using SIP, this feature is known as Shared
   Appearances of an AOR [BLISS-SHARED].  Extensions to the dialog
   package are used to convey appearance numbers (line numbers).

Appendix A.28.  Speakerphone Paging

   In Speakerphone Paging, Alice calls the paging address and speaks.
   Her voice is played on the speaker of every idle phone in a
   preconfigured group of phones.  Speakerphone paging can be
   implemented using either multicast or through a simple multipoint
   mixer.  In the multicast solution, the paging UA sends a multicast
   INVITE with send-only media in the SDP (see also [RFC3264]).  The
   automatic answer and enabling of the speakerphone is a locally
   configured decision on the paged UAs.  The paging UA sends RTP via
   the multicast address indicated in the SDP.

   The multipoint solution is accomplished by sending an INVITE to the
   multipoint mixer.  The mixer is configured to automatically answer
   the dialog.  The paging UA then sends REFER requests for each of the
   UAs that are to become paging speakers (the UA is likely to send out
   a single REFER that is parallel forked by the proxy server).  The UAs
   performing as paging speakers are configured to automatically answer
   based upon caller identification (e.g., the To field, URI, or
   Referred-To header fields).

   Finally, as a third option, the user agent can send a mass-invitation
   request to a conference server, which would create a conference and
   send INVITEs containing the Answer-Mode: Auto header field to all
   user agents in the paging group.

Appendix A.29.  Speed Dial

   In Speed Dial, Alice dials an abbreviated number, enters an alias, or
   presses a special speed-dial button representing Bob.  Her action is
   interpreted as if she specified the full address of Bob.

Appendix A.30.  Voice Message Screening

   In Voice Message Screening, Bob calls Alice.  Alice is screening her
   calls, so Bob hears Alice's voicemail greeting.  Alice can hear Bob
   leave his message.  If she decides to talk to Bob, she can take the
   call back from the voicemail system; otherwise, she can let Bob leave
   a message.  This emulates the behavior of a home telephone answering

Mahy, et al.                  Informational                    [Page 38]

RFC 5850               SIP Call Control Framework               May 2010

   At first, this is the same as Call Monitoring (Appendix A.7).  In
   this case, the voicemail service is one of the UAs.  The UA screening
   the message monitors the call on the voicemail service, and also
   subscribes to dialog information.  If the user screening their
   messages decides to answer, they perform a take from the voicemail
   system (for example, send an INVITE with Replaces to the UA leaving
   the message).

Appendix A.31.  Voice Portal

   Voice Portal is service that allows users to access a portal site
   using spoken dialog interaction.  For example, Alice needs to
   schedule a working dinner with her co-worker Carol.  Alice uses a
   voice portal to check Carol's flight schedule, find a restaurant near
   her hotel, make a reservation, get directions there, and page Carol
   with this information.  A voice portal is essentially a complex
   collection of voice dialogs used to access interesting content.  One
   of the most desirable call control features of a Voice Portal is the
   ability to start a new outgoing call from within the context of the
   Portal (to make a restaurant reservation, or return a voicemail
   message, for example).  Once the new call is over, the user should be
   able to return to the Portal by pressing a special key, using some
   DTMF sequence (e.g., a very long pound or hash tone), or by speaking
   a key word (e.g., "Main Menu").

   In order to accomplish this, the Voice Portal starts with the
   following media relationship:

   { User , Voice Portal }

   The user then asks to make an outgoing call.  The Voice Portal asks
   the user to perform a far-fork.  In other words, the Voice Portal
   wants the following media relationship:

           { Target , User }  &  { User , Voice Portal }

   The Voice Portal is now just listening for a key word or the
   appropriate DTMF.  As soon as the user indicates they are done, the
   Voice Portal takes the call from the old target, and we are back to
   the original media relationship.

   This feature can also be used by the account number and phone number
   collection menu in a pre-paid calling service.  A user can press a
   DTMF sequence that presents them with the appropriate menu again.

Mahy, et al.                  Informational                    [Page 39]

RFC 5850               SIP Call Control Framework               May 2010

Appendix A.32.  Voicemail

   In Voicemail, Alice calls Bob who does not answer or is not
   available.  The call forwards to a voicemail server that plays Bob's
   greeting and records Alice's message for Bob.  An indication is sent
   to Bob that a new message is waiting, and he retrieves the message at
   a later date.  This feature is implemented using features such as
   Call Forwarding (Appendix A.6) and the History-Info header field
   [RFC4244] or voicemail URI convention [RFC4458] and Message Waiting
   [RFC3842] features.

Appendix A.33.  Whispered Call Waiting

   In Whispered Call Waiting, Alice is in a conversation with Bob.
   Carol calls Alice.  Either Carol can "whisper" to Alice directly
   ("Can you get lunch in 15 minutes?"), or an automaton whispers to
   Alice informing her that Carol is trying to reach her.

Appendix B.  Acknowledgments

   The authors would like to acknowledge Ben Campbell for his
   contributions to the document and thank AC Mahendran, John Elwell,
   and Xavier Marjou for their detailed Working-Group review of the
   document.  The authors would like to thank Magnus Nystrom for his
   review of the document.

5.  Informative References

   [RFC3261]        Rosenberg, J., Schulzrinne, H., Camarillo, G.,
                    Johnston, A., Peterson, J., Sparks, R., Handley, M.,
                    and E. Schooler, "SIP: Session Initiation Protocol",
                    RFC 3261, June 2002.

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

   [RFC3265]        Roach, A., "Session Initiation Protocol (SIP)-
                    Specific Event Notification", RFC 3265, June 2002.

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

   [RFC5359]        Johnston, A., Sparks, R., Cunningham, C., Donovan,
                    S., and K. Summers, "Session Initiation Protocol
                    Service Examples", BCP 144, RFC 5359, October 2008.

Mahy, et al.                  Informational                    [Page 40]

RFC 5850               SIP Call Control Framework               May 2010

   [RFC3725]        Rosenberg, J., Peterson, J., Schulzrinne, H., and G.
                    Camarillo, "Best Current Practices for Third Party
                    Call Control (3pcc) in the Session Initiation
                    Protocol (SIP)", BCP 85, RFC 3725, April 2004.

   [RFC3515]        Sparks, R., "The Session Initiation Protocol (SIP)
                    Refer Method", RFC 3515, April 2003.

   [RFC3891]        Mahy, R., Biggs, B., and R. Dean, "The Session
                    Initiation Protocol (SIP) "Replaces" Header",
                    RFC 3891, September 2004.

   [RFC3911]        Mahy, R. and D. Petrie, "The Session Initiation
                    Protocol (SIP) "Join" Header", RFC 3911,
                    October 2004.

   [BLISS-PROBLEM]  Rosenberg, J., "Basic Level of Interoperability for
                    Session Initiation Protocol (SIP)  Services (BLISS)
                    Problem Statement", Work in Progress, March 2009.

   [RFC4235]        Rosenberg, J., Schulzrinne, H., and R. Mahy, "An
                    INVITE-Initiated Dialog Event Package for the
                    Session Initiation Protocol (SIP)", RFC 4235,
                    November 2005.

   [RFC4575]        Rosenberg, J., Schulzrinne, H., and O. Levin, "A
                    Session Initiation Protocol (SIP) Event Package for
                    Conference State", RFC 4575, August 2006.

   [RFC3680]        Rosenberg, J., "A Session Initiation Protocol (SIP)
                    Event Package for Registrations", RFC 3680,
                    March 2004.

   [RFC3856]        Rosenberg, J., "A Presence Event Package for the
                    Session Initiation Protocol (SIP)", RFC 3856,
                    August 2004.

   [RFC4353]        Rosenberg, J., "A Framework for Conferencing with
                    the Session Initiation Protocol (SIP)", RFC 4353,
                    February 2006.

   [RFC5629]        Rosenberg, J., "A Framework for Application
                    Interaction in the Session Initiation Protocol
                    (SIP)", RFC 5629, October 2009.

   [RFC5369]        Camarillo, G., "Framework for Transcoding with the
                    Session Initiation Protocol (SIP)", RFC 5369,
                    October 2008.

Mahy, et al.                  Informational                    [Page 41]

RFC 5850               SIP Call Control Framework               May 2010

   [XCON-CCMP]      Barnes, M., Boulton, C., Romano, S., and H.
                    Schulzrinne, "Centralized Conferencing Manipulation
                    Protocol", Work in Progress, February 2010.

   [RFC5589]        Sparks, R., Johnston, A., and D. Petrie, "Session
                    Initiation Protocol (SIP) Call Control - Transfer",
                    BCP 149, RFC 5589, June 2009.

   [RFC4579]        Johnston, A. and O. Levin, "Session Initiation
                    Protocol (SIP) Call Control - Conferencing for User
                    Agents", BCP 119, RFC 4579, August 2006.

   [RFC3840]        Rosenberg, J., Schulzrinne, H., and P. Kyzivat,
                    "Indicating User Agent Capabilities in the Session
                    Initiation Protocol (SIP)", RFC 3840, August 2004.

   [RFC3841]        Rosenberg, J., Schulzrinne, H., and P. Kyzivat,
                    "Caller Preferences for the Session Initiation
                    Protocol (SIP)", RFC 3841, August 2004.

   [RFC3087]        Campbell, B. and R. Sparks, "Control of Service
                    Context using SIP Request-URI", RFC 3087,
                    April 2001.

   [FEATURE-REF]    Audet, F., Johnston, A., Mahy, R., and C. Jennings,
                    "Feature Referral in the Session Initiation Protocol
                    (SIP)", Work in Progress, February 2008.

   [RFC4240]        Burger, E., Van Dyke, J., and A. Spitzer, "Basic
                    Network Media Services with SIP", RFC 4240,
                    December 2005.

   [RFC4458]        Jennings, C., Audet, F., and J. Elwell, "Session
                    Initiation Protocol (SIP) URIs for Applications such
                    as Voicemail and Interactive Voice Response (IVR)",
                    RFC 4458, April 2006.

   [RFC4538]        Rosenberg, J., "Request Authorization through Dialog
                    Identification in the Session Initiation Protocol
                    (SIP)", RFC 4538, June 2006.

   [RFC3880]        Lennox, J., Wu, X., and H. Schulzrinne, "Call
                    Processing Language (CPL): A Language for User
                    Control of Internet Telephony Services", RFC 3880,
                    October 2004.

Mahy, et al.                  Informational                    [Page 42]

RFC 5850               SIP Call Control Framework               May 2010

   [RFC5373]        Willis, D. and A. Allen, "Requesting Answering Modes
                    for the Session Initiation Protocol (SIP)",
                    RFC 5373, November 2008.

   [RFC3842]        Mahy, R., "A Message Summary and Message Waiting
                    Indication Event Package for the Session Initiation
                    Protocol (SIP)", RFC 3842, August 2004.

   [BLISS-SHARED]   Johnston, A., Soroushnejad, M., and V.
                    Venkataramanan, "Shared Appearances of a Session
                    Initiation Protocol (SIP) Address of Record (AOR)",
                    Work in Progress, October 2009.

   [RFC4244]        Barnes, M., "An Extension to the Session Initiation
                    Protocol (SIP) for Request History Information",
                    RFC 4244, November 2005.

   [RFC4313]        Oran, D., "Requirements for Distributed Control of
                    Automatic Speech Recognition (ASR), Speaker
                    Identification/Speaker Verification (SI/SV), and
                    Text-to-Speech (TTS) Resources", RFC 4313,
                    December 2005.

   [RFC3323]        Peterson, J., "A Privacy Mechanism for the Session
                    Initiation Protocol (SIP)", RFC 3323, November 2002.

   [RFC3325]        Jennings, C., Peterson, J., and M. Watson, "Private
                    Extensions to the Session Initiation Protocol (SIP)
                    for Asserted Identity within Trusted Networks",
                    RFC 3325, November 2002.

Mahy, et al.                  Informational                    [Page 43]

RFC 5850               SIP Call Control Framework               May 2010

Authors' Addresses

   Rohan Mahy

   EMail: rohan@ekabal.com

   Robert Sparks

   EMail: rjsparks@nostrum.com

   Jonathan Rosenberg

   EMail: jdrosen@jdrosen.net

   Dan Petrie

   EMail: dan.ietf@sipez.com

   Alan Johnston (editor)

   EMail: alan.b.johnston@gmail.com

Mahy, et al.                  Informational                    [Page 44]