RFC 2848
The PINT Service Protocol: Extensions to SIP and SDP for IP Access to Telephone Call Services
Pages: 73
Proposed Standard
Proposed Standard
Part 2 of 4 – Pages 8 to 37
3. PINT Functional and Protocol Architecture
3.1. PINT Functional Architecture
Familiarity is assumed with SIP 2.0 [1] and with SDP [2]. PINT clients and servers are SIP clients and servers. SIP is used to carry the request over the IP network to the correct PINT server in a secure and reliable manner, and SDP is used to describe the telephone network session that is to be invoked or whose status is to be returned. A PINT system uses SIP proxy servers and redirect servers for their usual purpose, but at some point there must be a PINT server with the means to relay received requests into a telephone system and to receive acknowledgement of these relayed requests. A PINT server with this capability is called a "PINT gateway". A PINT gateway appears to a SIP system as a User Agent Server. Notice that a PINT gateway appears to the PINT infrastructure as if it represents a "user", while in fact it really represents an entire telephone network infrastructure that can provide a set of telephone network services.
So the PINT system might appear to an individual PINT client as
follows:
/\/\/\/\/\/\/\ /\/\/\/\/\/\/\/\
___________ \ __/___ ___\_ \
| PINT | PINT \ PINT | PINT | |Exec| Telephone /
| client |<-------------->| server |gatewy|=====|Syst| Network \
|_________| protocol / cloud |______| |____| Cloud /
\ \ / \
/\/\/\/\/\/\/\ \/\/\/\/\/\/\/\/
Figure 1: PINT Functional Architecture
The system of PINT servers is represented as a cloud to emphasise
that a single PINT request might pass through a series of location
servers, proxy servers, and redirect servers, before finally reaching
the correct PINT gateway that can actually process the request by
passing it to the Telephone Network Cloud.
The PINT gateway might have a true telephone network interface, or it
might be connected via some other protocol or API to an "Executive
System" that is capable of invoking services within the telephone
cloud.
As an example, within an I.N. (Intelligent Network) system, the PINT
gateway might appear to realise the Service Control Gateway Function.
In an office environment, it might be a server adjunct to the office
PBX, connected to both the office LAN and the office PBX.
The Executive System that lies beyond the PINT gateway is outside the
scope of PINT.
3.2. PINT Protocol Architecture
This section explains how SIP and SDP work in combination to convey
the information necessary to invoke telephone network sessions.
The following list summarises the extension features used in PINT
1.0. Following on from this the features are considered separately
for SDP and then for SIP:
1) Telephony URLs in SDP Contact Fields
2) Refinement of SIP/SDP Telephony URLs
* Inclusion of private dialling plans
3) Specification of Telephone Service Provider (TSP) and/or phone-
context URL-parameters
4) Data Objects as session media
4a) Protocol Transport formats to indicate the treatment of the media
within the GSTN
5) Implicit (Indirect) media streams and opaque arguments
6) In-line data objects using multipart/mime
7) Refinement/Clarification of Opaque arguments passed onwards to
Executive Systems
* Framework for Presentation Restriction Indication
* Framework for Q.763 arguments
8) An extension mechanism for SDP to specify strictures and force
failure when a recipient does NOT support the specified
extensions, using "require" headers.
9) Mandatory support for "Warning" headers to give more detailed
information on request disposition.
10) Mechanism to register interest in the disposition of a requested
service, and to receive indications on that disposition.
Both PINT and SIP rely on features of MIME[4]. The use of SIP 2.0 is
implied by PINT 1.0, and this also implies compliance with version
1.0 of MIME.
3.2.1. SDP operation in PINT
The SDP payload contains a description of the particular telephone
network session that the requestor wishes to occur in the GSTN. This
information includes such things as the telephone network address
(i.e. the "telephone number") of the terminal(s) involved in the
call, an indication of the media type to be transported (e.g. audio,
text, image or application data), and an indication if the
information is to be transported over the telephone network via
voice, fax, or pager transport. An indication of the content to be
sent to the remote telephone terminal (if there is any) is also
included.
SDP is flexible enough to convey these parameters independently. For
example, a request to send some text via voice transport will be
fulfilled by invoking some text-to-speech-over-the-phone service, and
a request to send text via fax will be fulfilled by invoking some
text-to-fax service.
The following is a list of PINT 1.0 enhancements and additions to
SDP.
a. A new network type "TN" and address types "RFC2543" and "X-..."
(section 3.4.1)
b. New media types "text", "image", and "application", new
protocol transport keywords "voice", "fax" and "pager" and the
associated format types and attribute tags (section 3.4.2)
c. New format specific attributes for included content data
(section 3.4.2.4)
d. New attribute tags, used to pass information to the telephone
network (section 3.4.3)
e. A new attribute tag "require", used by a client to indicate
that some attribute is required to be supported in the server
(section 3.4.4)
3.2.2. SIP Operation in PINT
SIP is used to carry the request for telephone service from the PINT
client to the PINT gateway, and may include a telephone number if
needed for the particular service. The following is a complete list
of PINT enhancements and additions to SIP:
f. The multipart MIME payloads (section 3.5.1)
g. Mandatory support for "Warning:" headers (section 3.5.2)
h. The SUBSCRIBE and NOTIFY, and UNSUBSCRIBE requests (section
3.5.3)
i. Require: headers (section 3.5.4)
j. A format for PINT URLS within a PINT request (section 3.5.5)
k. Telephone Network Parameters within PINT URLs (section 3.5.6)
Section 3.5.8 contains remarks about how BYE requests are used within
PINT. This is not an extension to baseline SIP; it is included here
only for clarification of the semantics when used with telephone
network sessions.
3.3. REQUIRED and OPTIONAL elements for PINT compliance
Of these, only the TN network type (with its associated RFC2543
address type) and the "require" attribute MUST be supported by PINT
1.0 clients and servers. In practice, most PINT service requests will
use other changes, of which references to Data Objects in requests
are most likely to appear in PINT requests.
Each of the other new PINT constructs enables a different function,
and a client or server that wishes to enable that particular function
MUST do so by the construct specified in this document. For example,
building a PINT client and server that provide only the Request-to-
Call telephone call service, without support for the other Milestone
services, is allowed.
The "Require:" SIP header and the "require" attribute provide a
mechanism that can be used by clients and servers to signal their
need and/or ability to support specific "new" PINT protocol elements.
It should be noted that many optional features of SIP and SDP make sense as specified in the PINT context. One example is the SDP a=lang: attribute, which can be used to describe the preferred language of the callee. Another example is the use of the "t=" parameter to indicate that the time at which the PINT service is to be invoked. This is the normal use of the "t=" field. A third example is the quality attributes. Any SIP or SDP option or facility is available to PINT clients and servers without change. Conversely, support for Data Objects within Internet Conference sessions may be useful, even if the aim is not to provide a GSTN service request. In this case, the extensions covering these items may be incorporated into an otherwise "plain" SIP/SDP invitation. Likewise, support for SDP "require" may be useful, as a framework for addition of features to a "traditional" SIP/SDP infrastructure. Again, these may be convenient to incorporate into SIP/SDP implementations that would not be used for PINT service requests. Such additions are beyond the scope of this document, however.3.4. PINT Extensions to SDP
PINT 1.0 adds to SDP the possibility to describe audio, fax, and pager telephone sessions. It is deliberately designed to hide the underlying technical details and complexity of the telephone network. The only network type defined for PINT is the generic "TN" (Telephone Network). More precise tags such as "ISDN", "GSM", are not defined. Similarly, the transport protocols are designated simply as "fax", "voice", and "pager"; there are no more specific identifiers for the various telephone network voice, fax, or pager protocols. Similarly, the data to be transported are identified only by a MIME content type, such as "text" data, "image" data, or some more general "application" data. An important example of transporting "application" data is the milestone service "Voice Access to Web Content". In this case the data to be transported are pointed to by a URI, the data content type is application/URI, and the transport protocol would be "voice". Some sort of speech-synthesis facility, speaking out to a Phone, will have to be invoked to perform this service. This section gives details of the new SDP keywords.3.4.1. Network Type "TN" and Address Type "RFC2543"
The TN ("Telephone Network") network type is used to indicate that the terminal is connected to a telephone network. The address types allowed for network type TN are "RFC2543" and private address types, which MUST begin with an "X-".
Address type RFC2543 is followed by a string conforming to a subset of the "telephone-subscriber" BNF specified in figure 4 of SIP [1]). Note that this BNF is NOT identical to the BNF that defines the "phone-number" within the "p=" field of SDP. Examples: c= TN RFC2543 +1-201-406-4090 c= TN RFC2543 12014064090 A telephone-subscriber string is of one of two types: global-phone- number or local-phone-number. These are distinguished by preceeding a global-phone-number with a "plus" sign ("+"). A global-phone-number is by default to be interpreted as an internationally significant E.164 Number Plan Address, as defined by [6], whilst a local-phone- number is a number specified in the default dialling plan within the context of the recipient PINT Gateway. An implementation MAY use private addressing types, which can be useful within a local domain. These address types MUST begin with an "X-", and SHOULD contain a domain name after the X-, e.g. "X- mytype.mydomain.com". An example of such a connection line is as follows: c= TN X-mytype.mydomain.com A*8-HELEN where "X-mytype.mydomain.com" identifies this private address type, and "A*8-HELEN" is the number in this format. Such a format is defined as an "OtherAddr" in the ABNF of Appendix A. Note that most dialable telephone numbers are expressable as local-phone-numbers within address RFC2543; new address types SHOULD only be used for formats which cannot be so written.3.4.2. Support for Data Objects within PINT
One significant change over traditional SIP/SDP Internet Conference sessions with PINT is that a PINT service request may refer to a Data Object to be used as source information in that request. For example, a PINT service request may specify a document to be processed as part of a GSTN service by which a Fax is sent. Similarly, a GSTN service may be take a Web page and result in a vocoder processing that page and speaking the contents over a telephone. The SDP specification does not have explicit support for reference to or carriage of Data Objects within requests. In order to use SDP for PINT, there is a need to describe such media sessions as "a telephone
call to a certain number during which such-and-such an image is sent as a fax". To support this, two extensions to the session description format are specified. These are some new allowed values for the Media Field, and a description of the "fmtp" parameter when used with the Media Field values (within the context of the Contact Field Network type "TN"). An addition is also made to the SIP message format to allow the inclusion of data objects as sub-parts within the request message itself. The original SDP syntax (from [2]) for media-field is given as: media-field = "m=" media space port ["/" integer] space proto 1*(space fmt) CRLF When used within PINT requests, the definition of the sub-fields is expanded slightly. The Media sub-field definition is relaxed to accept all of the discrete "top-level" media types defined in [4]. In the milestone services the discrete type "video" is not used, and the extra types "data" and "control" are likewise not needed. The use of these types is not precluded, but the behaviour expected of a PINT Gateway receiving a request including such a type is not defined here. The Port sub-field has no meaning in PINT requests as the destination terminals are specified using "TN" addressing, so the value of the port sub-field in PINT requests is normally set to "1". A value of "0" may be used as in SDP to indicate that the terminal is not receiving media. This is useful to indicate that a telephone terminal has gone "on hold" temporarily. Likewise, the optional integer sub-field is not used in PINT. As mentioned in [2], the Transport Protocol sub-field is specific to the associated Address Type. In the case that the Address Type in the preceeding Contact field is one of those defined for use with the Network Type "TN", the following values are defined for the Transport Protocol sub-field: "voice", "fax", and "pager". The interpretation of this sub-field within PINT requests is the treatment or disposition of the resulting GSTN service. Thus, for transport protocol "voice", the intent is that the service will result in a GSTN voice call, whilst for protocol "fax" the result will be a GSTN fax transmission, and protocol "pager" will result in a pager message being sent.
Note that this sub-field does not necessarily dictate the media type and subtype of any source data; for example, one of the milestone services calls for a textual source to be vocoded and spoken in a resulting telephone service call. The transport protocol value in this case would be "voice", whilst the media type would be "text". The Fmt sub-field is described in [2] as being transport protocol- specific. When used within PINT requests having one of the above protocol values, this sub-field consists of a list of one or more values, each of which is a defined MIME sub-type of the associated Media sub-field value. The special value "-" is allowed, meaning that there is no MIME sub-type. This sub-field retains (from [2]) its meaning that the list will contain a set of alternative sub-types, with the first being the preferred value. For experimental purposes and by mutual consent of the sender and recipient, a sub-type value may be specified as an <X-token>, i.e. a character string starting with "X-". The use of such values is discouraged, and if such a value is expected to find common use then it SHOULD be registered with IANA using the standard content type registration process (see Appendix C). When the Fmt parameter is the single character "-" ( a dash ), this is interpreted as meaning that a unspecified or default sub-type can be used for this service. Thus, the media field value "m=audio 1 voice -<CRLF>" is taken to mean that a voice call is requested, using whatever audio sub type is deemed appropriate by the Executive System. PINT service is a special case, in that the request comes from the IP network but the service call is provided within the GSTN. Thus the service request will not normally be able to define the particular codec used for the resulting GSTN service call. If such an intent IS required, then the quality attribute may be used (see "Suggested Attributes" section of [2]).3.4.2.1. Use of fmtp attributes in PINT requests
For each element of the Fmt sub-field, there MUST be a following fmtp attribute. When used within PINT requests, the fmtp attribute has a general structure as defined here: "a=fmtp:" <subtype> <space> resolution *(<space> resolution) (<space> ";" 1(<attribute>) *(<space> <attribute>)) where: <resolution> := (<uri-ref> | <opaque-ref> | <sub-part-ref>)
A fmtp attribute describes the sources used with a given Fmt entry in the Media field. The entries in a Fmt sub-field are alternatives (with the preferred one first in the list). Each entry will have a matching fmtp attribute. The list of resolutions in a fmtp attribute describes the set of sources that resolve the matching Fmt choice; all elements of this set will be used. It should be noted that, for use in PINT services, the elements in such a set will be sent as a sequence; it is unlikely that trying to send them in parallel would be successful. A fmtp attribute can contain a mixture of different kinds of element. Thus an attribute might contain a sub-part-ref indicating included data held in a sub-part of the current message, followed by an opaque-ref referring to some content on the GSTN, followed by a uri- ref pointing to some data held externally on the IP network. To indicate which form each resolution element takes, each of them starts with its own literal tag. The detailed syntax of each form is described in the following sub-sections.3.4.2.2. Support for Remote Data Object References in PINT
Where data objects stored elsewhere on the IP Network are to be used as sources for processing within a PINT service, they may be referred to using the uri-ref form. This is simply a Uniform Resource Identifier (URI), as described in [9]. Note that the reference SHOULD be an absolute URI, as there may not be enough contextual information for the recipient server to resolve a relative reference; any use of relative references requires some private agreement between the sender and recipient of the message, and SHOULD be avoided unless the sender can be sure that the recipient is the one intended and the reference is unambiguous in context. This also holds for partial URIs (such as"uri:http://aNode/index.htm") as these will need to be resolved in the context of the eventual recipient of the message. The general syntax of a reference to an Internet-based external data object in a fmtp line within a PINT session description is: <uri-ref> := ("uri:" URI-reference) where URI-reference is as defined in Appendix A of [9]
For example:
c= TN RFC2543 +1-201-406-4090
m= text 1 fax plain
a=fmtp:plain uri:ftp://ftp.isi.edu/in-notes/rfc2468.txt
or:
c= TN RFC2543 +1-201-406-4090
m= text 1 fax plain
a=fmtp:plain
uri:http://www.ietf.org/meetings/glance_minneapolis.txt
means get this data object from the Internet and use it as a source
for the requested GSTN Fax service.
3.4.2.3. Support for GSTN-based Data Objects in PINT
PINT services may refer to data that are held not on the IP Network
but instead within the GSTN. The way in which these items are
indicated need have no meaning within the context of the Requestor or
the PINT Gateway; the reference is merely some data that may be used
by the Executive System to indicate the content intended as part of
the request. These data form an opaque reference, in that they are
sent "untouched" through the PINT infrastructure.
A reference to some data object held on the GSTN has the general
definition:
<opaque-ref> := ("opr:" *uric)
where uric is as defined in Appendix A of [9].
For example:
c= TN RFC2543 +1-201-406-4090
m= text 1 fax plain
a=fmtp:plain opr:APPL.123.456
means send the data that is indexed ON THE GSTN by the reference
value "APPL.123.456" to the fax machine on +1-201-406-4090. The
Executive System may also take the Telephone URL held in the To:
field of the enclosing SIP message into account when deciding the
context to be used for the data object dereference.
Of course, an opaque reference may also be used for other purposes;
it could, for example, be needed to authorise access to a document
held on the GSTN rather than being required merely to disambiguate
the data object. The purpose to which an opaque reference is put,
however, is out of scope for this document. It is merely an indicator
carried within a PINT Request.
An opaque reference may have no value in the case where the value to
be used is implicit in the rest of the request. For example, suppose
some company wishes to use PINT to implement a "fax-back service". In
their current implementation, the image(s) to be faxed are entirely
defined by the telephone number dialled. Within the PINT request,
this telephone number would appear within the "To:" field of the PINT
request, and so there is no need for an opaque reference value.
If there are several resolutions for a PINT Service Request, and one
of these is an opaque reference with no value, then that opaque
reference MUST be included in the attribute line, but with an empty
value field.
For example:
c= TN RFC2543 +1-201-406-4090
m= text 1 fax plain
a=fmtp:plain uri:http://www.sun.com/index.html opr:
might be used to precede some data to be faxed with a covering note.
In the special case where an opaque reference is the sole resolution
of a PINT Service Request, AND that reference needs no value, there
is no need for a Fmt list at all; the intent of the service is
unambiguous without any further resolution.
For example:
c= TN RFC2543 +1-201-406-4090
m= text 1 fax -
means that there is an implied content stored on the GSTN, and that
this is uniquely identified by the combination of SIP To-URI and the
Contact field of the session description.
3.4.2.4. Session Description support for included Data Objects
As an alternative to pointing to the data via a URI or an opaque
reference to a data item held on the GSTN, it is possible to include
the content data within the SIP request itself. This is done by using
multipart MIME for the SIP payload. The first MIME part contains the
SDP description of the telephone network session to be executed. The
other MIME parts contain the content data to be transported.
Format specific attribute lines within the session description are
used to indicate which other MIME part within the request contains
the content data. Instead of a URI or opaque reference, the format-
specific attribute indicates the Content-ID of the MIME part of the
request that contains the actual data, and is defined as:
<sub-part-ref> := ("spr:" Content-ID)
where Content-ID is as defined in Appendix A of [3] and in [10]).
For example:
c= TN RFC2543 +1-201-406-4090
m= text 1 fax plain
a=fmtp:plain spr:<Content-ID>
The <Content-ID> parameter is the Content-ID of one of the MIME parts
inside the message, and this fragment means that the requesting user
would like the data object held in the sub-part of this message
labelled <Content-ID> to be faxed to the machine at phone number +1-
201-406-4090.
See also section 3.5.1 for a discussion on the support needed in the
enclosing SIP request for included data objects.
3.4.3. Attribute Tags to pass information into the Telephone Network
It may be desired to include within the PINT request service
parameters that can be understood only by some entity in the
"Telephone Network Cloud". SDP attribute parameters are used for this
purpose. They MAY appear within a particular media description or
outside of a media description.
These attributes may also appear as parameters within PINT URLS (see
section 3.5.6) as part of a SIP request.
This is necessary so that telephone terminals that require the
attributes to be defined can appear within the To: line of a PINT
request as well as within PINT session descriptions.
The purpose of these attributes is to allow the client to specify
extra context within which a particular telephone number is to be
interpreted. There are many reasons why extra context might be
necessary to interpret a given telephone number:
a. The telephone number might be reachable in many different ways
(such as via competing telephone service providers), and the
PINT client wishes to indicate its selection of service
provider.
b. The telephone number might be reachable only from a limited
number of networks (such as an '800' freephone number).
c. The telephone number might be reachable only within a single
telephone network (such as the '152' customer service number of
BT). Similarly, the number might be an internal corporate
extension reachable only within the PBX.
However, as noted above, it is not usually necessary to use SDP
attributes to specify the phone context. URLs such as
152@pint.bt.co.il within the To: and From: headers and/or Request-
URI, normally offer sufficient context to resolve telephone numbers.
If the client wishes the request to fail if the attributes are not
supported, these attributes SHOULD be used in conjunction with the
"require" attribute (section 3.4.4) and the
"Require:org.ietf.sdp.require" header (section 3.5.4).
It is not possible to standardise every possible internal telephone
network parameter. PINT 1.0 attributes have been chosen for
specification because they are common enough that many different PINT
systems will want to use them, and therefore interoperability will be
increased by having a single specification.
Proprietary attribute "a=" lines, that by definition are not
interoperable, may be nonetheless useful when it is necessary to
transport some proprietary internal telephone network variables over
the IP network, for example to identify the order in which service
call legs are to be be made. These private attributes SHOULD BE,
however, subject to the same IANA registration procedures mentioned
in the SDP specification[2] (see also this Appendix C).
3.4.3.1. The phone-context attribute
An attribute is specified to enable "remote local dialling". This is
the service that allows a PINT client to reach a number from far
outside the area or network that can usually reach the number. It is
useful when the sending or receiving address is only dialable within
some local context, which may be remote to the origin of the PINT
client.
For example, if Alice wanted to report a problem with her telephone,
she might then dial a "network wide" customer care number; within the
British Telecom network in the U.K., this is "152". Note that in this
case she doesn't dial any trunk prefix - this is the whole dialable
number. If dialled from another operator's network, it will not
connect to British Telecom's Engineering Enquiries service; and
dialling "+44 152" will not normally succeed. Such numbers are called
Network-Specific Service Numbers.
Within the telephone network, the "local context" is provided by the
physical connection between the subscriber's terminal and the central
office. An analogous association between the PINT client and the PINT
server that first receives the request may not exist, which is why it
may be necessary to supply this missing "telephone network context".
This attribute is defined as follows:
a=phone-context: <phone-context-ident>
phone-context-ident = network-prefix / private-prefix
network-prefix = intl-network-prefix / local-network-prefix
intl-network-prefix = "+" 1*DIGIT
local-network-prefix = 1*DIGIT
excldigandplus = (0x21-0x2d,0x2f,0x40-0x7d))
private-prefix = 1*excldigandplus 0*uric
An intl-network-prefix and local-network-prefix MUST be a bona fide
network prefix, and a network-prefix that is an intl-network-prefix
MUST begin with an E.164 service code ("country code").
It is possible to register new private-prefixes with IANA so as to
avoid collisions. Prefixes that are not so registered MUST begin with
an "X-" to indicate their private, non-standard nature (see Appendix
C).
Example 1:
c= TN RFC2543 1-800-765-4321
a=phone-context:+972
This describes an terminal whose address in Israel (E.164 country
code 972) is 1-800-765-4321.
Example 2:
c= TN RFC2543 1-800-765-4321
a=phone-context:+1
This describes an terminal whose address in North America (E.164
country code 1) is 1-800-765-4321.
The two telephone terminals described by examples 1 and 2 are
different; in fact they are located in different countries.
Example 3:
c=TN RFC2543 123
a=phone-context:+97252
This describes a terminal whose address when dialled from within the
network identified by +97252 is the string "123". It so happens that
+97252 defines one of the Israeli cell phone providers, and 123
reaches customer service when dialled within that network.
It may well be useful or necessary to use the SDP "require" parameter
in conjunction with the phone-context attribute.
Example 4:
c= TN RFC2543 321
a=phone-context:X-acme.com-23
This might describe the telephone terminal that is at extension 321
of PBX number 23 within the acme.com private PBX network. It is
expected that such a description would be understandable by the
acme.com PINT server that receives the request.
Note that if the PINT server receiving the request is inside the
acme.com network, the same terminal might be addressable as follows:
c= TN RFC2543 7-23-321
(assuming that "7" is dialled in order to reach the private PBX
network from within acme.com)
3.4.3.2. Presentation Restriction attribute
Although it has no affect on the transport of the service request
through the IP Network, there may be a requirement to allow
originators of a PINT service request to indicate whether or not they
wish the "B party" in the resulting service call to be presented with
the "A party's" calling telephone number. It is a legal requirement
in some jurisdictions that a caller be able to select whether or not
their correspondent can find out the calling telephone number (using
Automatic Number Indication or Caller Display or Calling Line
Identity Presentation equipment). Thus an attribute may be needed to
indicate the originator's preference.
Whether or not the default behaviour of the Executive System is to
present or not present a party's telephone number to the
correspondent GSTN terminal is not specified, and it is not mandatory
in all territories for a PINT Gateway or Executive System to act on
this attribute. It is, however, defined here for use where there are
regulatory restrictions on GSTN operation, and in that case the
Executive System can use it to honour the originator's request.
The attribute is specified as follows:
a=clir:<"true" | "false">
This boolean value is needed within the attribute as it may be that
the GSTN address is, by default, set to NOT present its identity to
correspondents, and the originator wants to do so for this particular
call. It is in keeping with the aim of this attribute to allow the
originator to specify what treatment they want for the requested
service call.
The expected interpretation of this attribute is that, if it is
present and the value is "false" then the Calling Line Identity CAN
be presented to the correspondent terminal, whilst if it is "true"
then if possible the Executive System is requested to NOT present the
Calling Line Identity.
3.4.3.3. ITU-T CalledPartyAddress attributes parameters
These attributes correspond to fields that appear within the ITU-T
Q.763 "CalledPartyAddress" field (see [8] ,section 3.9). PINT clients
use these attributes in order to specify further parameters relating
to Terminal Addresses, in the case when the address indicates a
"local-phone-number". In the case that the PINT request contains a
reference to a GSTN terminal, the parameters may be required to
correctly identify that remote terminal.
The general form of this attribute is: "a=Q763-<token>((":" <value>)
|"")". Three of the possible elements and their use in SDP
attributes are described here. Where other Q763 elements are to be
used, then these should be the subject of further specification to
define the syntax of the attribute mapping. It is recommended that
any such specification maintains the value sets shown in Q.763.
The defined attributes are:
a=Q763-nature: - indicates the "nature of address indicator".
The value MAY be any number between 0 and 127.
The following values are specified:
"1" a subscriber number
"2" unknown
"3" a nationally significant number
"4" an internationally significant number
The values have been chosen to coincide with the values in Q.763.
Note that other values are possible, according to national rules or
future expansion of Q.763.
a=Q763-plan: - indicates the numbering plan to which the address
belongs. The value MAY be any number between 0
and 7. The following values are specified:
"1" Telephone numbering plan (ITU-T E.164)
"3" Data numbering plan (ITU-T X.121)
"4" Telex numbering plan (ITU-T F.69)
The values have been chosen to coincide with the values in Q.763.
Other values are allowed, according to national rules or future
expansion of Q.763.
a=Q763-INN - indicates if routing to the Internal Network Number
is allowed. The value MUST be ONE of:
"0" routing to internal network number allowed
"1" routing to internal network number not
allowed
The values have been chosen to coincide with the values in Q.763.
Note that it is possible to use a local-phone-number and indicate via
attributes that the number is in fact an internationally significant
E.164 number. Normally this SHOULD NOT be done; an internationally
significant E.164 number is indicated by using a "global-phone-
number" for the address string.
3.4.4. The "require" attribute
According to the SDP specification, a PINT server is allowed simply
to ignore attribute parameters that it does not understand. In order
to force a server to decline a request if it does not understand one
of the PINT attributes, a client SHOULD use the "require" attribute,
specified as follows:
a=require:<attribute-list>
where the attribute-list is a comma-separated list of attributes that
appear elsewhere in the session description.
In order to process the request successfully the PINT server must
BOTH understand the attribute AND ALSO fulfill the request implied by
the presence of the attribute, for each attribute appearing within
the attribute-list of the require attribute.
If the server does not recognise the attribute listed, the PINT server MUST return an error status code (such as 420 (Bad Extension) or 400 (Bad Request)), and SHOULD return suitable Warning: lines explaining the problem or an Unsupported: header containing the attribute it does not understand. If the server recognizes the attribute listed, but cannot fulfill the request implied by the presence of the attribute, the request MUST be rejected with a status code of (606 Not Acceptable), along with a suitable Unsupported: header or Warning: line. The "require" attribute may appear anywhere in the session description, and any number of times, but it MUST appear before the use of the attribute marked as required. Since the "require" attribute is itself an attribute, the SIP specification allows a server that does not understand the require attribute to ignore it. In order to ensure that the PINT server will comply with the "require" attribute, a PINT client SHOULD include a Require: header with the tag "org.ietf.sdp.require" (section 3.5.4) Note that the majority of the PINT extensions are "tagged" and these tags can be included in Require strictures. The exception is the use of phone numbers in SDP parts. However, these are defined as a new network and address type, so that a receiving SIP/SDP server should be able to detect whether or not it supports these forms. The default behaviour for any SDP recipient is that it will fail a PINT request if it does not recognise or support the TN and RFC2543 or X-token network and address types, as without the contents being recognised no media session could be created. Thus a separate stricture is not required in this case.3.5. PINT Extensions to SIP 2.0
PINT requests are SIP requests; Many of the specifications within this document merely explain how to use existing SIP facilities for the purposes of PINT.3.5.1. Multi-part MIME (sending data along with SIP request)
A PINT request can contain a payload which is multipart MIME. In this case the first part MUST contain an SDP session description that includes at least one of the format specific attribute tags for "included content data" specified above in section 3.4.3. Subsequent parts contain content data that may be transferred to the requested Telephone Call Service. As discussed earlier, within a single PINT request, some of the data MAY be pointed to by a URI within the request, and some of the data MAY be included within the request.
Where included data is carried within a PINT service request, the Content Type entity header of the enclosing SIP message MUST indicate this. To do so, the media type value within this entity header MUST be set to a value of "multipart". There is a content sub-type that is intended for situations like this in which sub-parts are to be handled together. This is the multipart/related type (defined in [19]), and it's use is recommended. The enclosed body parts SHOULD include the part-specific Content Type headers as appropriate ("application/sdp" for the first body part holding the session description, with an appropriate content type for each of the subsequent, "included data object" parts). This matches the standard syntax of MIME multipart messages as defined in [4]. For example, in a multipart message where the string "------next-------" is the boundary, the first two parts might be as follows: ------next------- Content-Type: application/sdp .... c= TN RFC2543 +1-201-406-4090 m= text 1 pager plain a=fmtp:plain spr:17@mymessage.acme.com ----------next------- Content-Type: text/plain Content-ID: 17@mymessage.acme.com This is the text that is to be paged to +1-201-406-4090 ----------next----------- The ability to indicate different alternatives for the content to be transported is useful, even when the alternatives are included within the request. For example, a request to send a short message to a pager might include the message in Unicode [5] and an alternative version of the same content in text/plain, should the PINT server or telephone network not be able to process the unicode. PINT clients should be extremely careful when sending included data within a PINT request. Such requests SHOULD be sent via TCP, to avoid fragmentation and to transmit the data reliably. It is possible that the PINT server is a proxy server that will replicate and fork the request, which could be disastrous if the request contains a large amount of application data. PINT proxy servers should be careful not to create many copies of a request with large amounts of data in it.
If the client does not know the actual location of the PINT gateway, and is using the SIP location services to find it, and the included data makes the PINT request likely to be transported in several IP datagrams, it is RECOMMENDED that the initial PINT request not include the data object but instead hold a reference to it.3.5.2. Warning header
A PINT server MUST support the SIP "Warning:" header so that it can signal lack of support for individual PINT features. As an example, suppose the PINT request is to send a jpeg picture to a fax machine, but the server cannot retrieve and/or translate jpeg pictures from the Internet into fax transmissions. In such a case the server fails the request and includes a Warning such as the following: Warning: 305 pint.acme.com Incompatible media format: jpeg SIP servers that do not understand the PINT extensions at all are strongly encouraged to implement Warning: headers to indicate that PINT extensions are not understood. Also, Warning: headers may be included within NOTIFY requests if it is necessary to notify the client about some condition concerning the invocation of the PINT service (see next).3.5.3. Mechanism to register interest in the disposition of a PINT service, and to receive indications on that disposition
It can be very useful to find out whether or not a requested service has completed, and if so whether or not it was successful. This is especially true for PINT service, where the person requesting the service is not (necessarily) a party to it, and so may not have an easy way of finding out the disposition of that service. Equally, it may be useful to indicate when the service has changed state, for example when the service call has started. Arranging a flexible system to provide extensive monitoring and control during a service is non-trivial (see section 6.4 for some issues); PINT 1.0 uses a simple scheme that should nevertheless provide useful information. It is possible to expand the scheme in a "backwards compatible" manner, so if required it can be enhanced at a later date. The PINT 1.0 status registration and indication scheme uses three new methods; SUBSCRIBE, UNSUBSCRIBE, and NOTIFY. These are used to allow a PINT client to register an interest in (or "subscribe" to) the
status of a service request, to indicate that a prior interest has lapsed (i.e "unsubscribe" from the status), and for the server to return service indications. The state machine of SUBSCRIBE/UNSUBSCRIBE is identical to that of INVITE/BYE; just as INVITE signals the beginning and BYE signals the end of participation in a media session, SUBSCRIBE signals the beginning and UNSUBSCRIBE signals the end of participation in a monitoring session. During the monitoring session, NOTIFY messages are sent to inform the subscriber of a change in session state or disposition.3.5.3.1. Opening a monitoring session with a SUBSCRIBE request
When a SUBSCRIBE request is sent to a PINT Server, it indicates that a user wishes to receive information about the status of a service session. The request identifies the session of interest by including the original session description along with the request, using the SDP global-session-id that forms part of the origin-field to identify the service session uniquely. The SUBSCRIBE request (like any other SIP request about an ongoing session) is sent to the same server as was sent the original INVITE, or to a server which was specified in the Contact: field within a subsequent response (this might well be the PINT gateway for the session). Whilst there are situations in which re-use of the Call-ID used in the original INVITE that initiated the session of interest is possible, there are other situations in which it is not. In detail, where the subscription is being made by the user who initiated the original service request, the Call-ID may be used as it will be known to the receiver to refer to a previously established session. However, when the request comes from a user other than the original requesting user, the SUBSCRIBE request constitutes a new SIP call leg, so the Call-ID SHOULD NOT be used; the only common identifier is the origin-field of the session description enclosed within the original service request, and so this MUST be used. Rather than have two different methods of identifying the "session of interest" the choice is to use the origin-field of the SDP sub-part included both in the original INVITE and in this SUBSCRIBE request. Note that the request MUST NOT include any sub-parts other than the session description, even if these others were present in the original INVITE request. A server MUST ignore whatever sub-parts are included within a SUBSCRIBE request with the sole exception of the enclosed session description.
The request MAY contain a "Contact:" header, specifying the PINT User Agent Server to which such information should be sent. In addition, it SHOULD contain an Expires: header, which indicates for how long the PINT Requestor wishes to receive notification of the session status. We refer to the period of time before the expiration of the SUBSCRIBE request as the "subscription period". See section 5.1.4. for security considerations, particularly privacy implications. A value of 0 within the Expires: header indicates a desire to receive one single immediate response (i.e. the request expires immediately). It is possible for a sequence of monitoring sessions to be opened, exist, and complete, all relating to the same service session. A successful response to the SUBSCRIBE request includes the session description, according to the Gateway. Normally this will be identical to the last cached response that the Gateway returned to any request concerning the same SDP global session id (see [2], section 6, o= field). The t= line may be altered to indicate the actual start or stop time, however. The Gateway might add an i= line to the session description to indicate such information as how many fax pages were sent. The Gateway SHOULD include an Expires: header indicating how long it is willing to maintain the monitoring session. If this is unacceptable to the PINT Requestor, then it can close the session by sending an immediate UNSUBSCRIBE message (see 3.5.3.3). In principle, a user might send a SUBSCRIBE request after the telephone network service has completed. This allows, for example, checking up "the morning after" to see if the fax was successfully transmitted. However, a PINT gateway is only required to keep state about a call for as long as it indicated previously in an Expires: header sent within the response to the original INVITE message that triggered the service session, within the response to the SUBSCRIBE message, within the response to any UNSUBSCRIBE message, or within its own UNSUBSCRIBE message (but see section 3.5.8, point 3). If the Server no longer has a record of the session to which a Requestor has SUBSCRIBEd, it returns "606 Not Acceptable", along with the appropriate Warning: 307 header indicating that the SDP session ID is no longer valid. This means that a requesting Client that knows that it will want information about the status of a session after the session terminates SHOULD send a SUBSCRIBE request before the session terminates.
3.5.3.2. Sending Status Indications with a NOTIFY request
During the subscription period, the Gateway may, from time to time, send a spontaneous NOTIFY request to the entity indicated in the Contact: header of the "opening" SUBSCRIBE request. Normally this will happen as a result of any change in the status of the service session for which the Requestor has subscribed. The receiving user agent server MUST acknowledge this by returning a final response (normally a "200 OK"). In this version of the PINT extensions, the Gateway is not required to support redirects (3xx codes), and so may treat them as a failure. Thus, if the response code class is above 2xx then this may be treated by the Gateway as a failure of the monitoring session, and in that situation it will immediately attempt to close the session (see next). The NOTIFY request contains the modified session description. For example, the Gateway may be able to indicate a more accurate start or stop time. The Gateway may include a Warning: header to describe some problem with the invocation of the service, and may indicate within an i= line some information about the telephone network session itself. Example: NOTIFY sip:petrack@pager.com SIP/2.0 To: sip:petrack@pager.com From: sip:R2F.pint.com@service.com Call-ID: 19971205T234505.56.78@pager.com CSeq: 4711 SUBSCRIBE Warning: xxx fax aborted, will try for the next hour. Content-Type:application/sdp c=... i=3 pages of 5 sent t=...3.5.3.3. Closing a monitoring session with an UNSUBSCRIBE request
At some point, either the Client's representative User Agent Server or the Gateway may decide to terminate the monitoring session. This is achieved by sending an UNSUBSCRIBE request to the correspondent server. Such a request indicates that the sender intends to close the monitoring session immediately, and, on receipt of the final response from the receiving server, the session is deemed over.
Note that unlike the SUBSCRIBE request, which is never sent by a PINT gateway, an UNSUBSCRIBE request can be sent by a PINT gateway to the User Agent Server to indicate that the monitoring session is closed. (This is analogous to the fact that a gateway never sends an INVITE, although it can send a BYE to indicate that a telephone call has ended.) If the Gateway initiates closure of the monitoring session by sending an UNSUBSCRIBE message, it SHOULD include an "Expires:" header showing for how much longer after this monitoring session is closed it is willing to store information on the service session. This acts as a minimum time within which the Client can send a new SUBSCRIBE message to open another monitoring session; after the time indicated in the Expires: header the Gateway is free to dispose of any record of the service session, so that subsequent SUBSCRIBE requests can be rejected with a "606" response. If the subscription period specified by the Client has expired, then the Gateway may send an immediate UNSUBSCRIBE request to the Client's representative User Agent Server. This ensures that the monitoring session always completes with a UNSUBSCRIBE/response exchange, and that the representative User Agent Server can avoid maintaining state in certain circumstances.3.5.3.4. Timing of SUBSCRIBE requests
As it relies on the Gateway having a copy of the INVITEd session description, the SUBSCRIBE message is limited in when it can be issued. The Gateway must have received the service request to which this monitoring session is to be associated, which from the Client's perspective happens as soon as the Gateway has sent a 1xx response back to it. However, once this has been done, there is no reason why the Client should not send a monitoring request. It does not have to wait for the final response from the Gateway, and it can certainly send the SUBSCRIBE request before sending the ACK for the Service request final response. Beyond this point, the Client is free to send a SUBSCRIBE request when it decides, unless the Gateway's final response to the initial service request indicated a short Expires: time. However, there are good reasons (see 6.4) why it may be appropriate to start a monitoring session immediately before the service is confirmed by the PINT Client sending an ACK. At this point the Gateway will have decided whether or not it can handle the service request, but will not have passed the request on to the Executive System. It is therefore in a good position to ask the Executive
System to enable monitoring when it sends the service request onwards. In practical implementations, it is likely that more information on transient service status will be available if this is indicated as being important BEFORE or AS the service execution phase starts; once execution has begun the level of information that can be returned may be difficult to change. Thus, whilst it is free to send a SUBSCRIBE request at any point after receiving an Interim response from the Gateway to its service request, it is recommended that the Client should send such a monitoring request immediately prior to sending an ACK message confirming the service if it is interested in transient service status messages.3.5.4. The "Require:" header for PINT
PINT clients use the Require: header to signal to the PINT server that a certain PINT extension of SIP is required. PINT 1.0 defines two strings that can go into the Require header: org.ietf.sip.subscribe -- the server can fulfill SUBSCRIBE requests and associated methods (see section 3.5.3) org.ietf.sdp.require -- the PINT server (or the SDP parser associated to it) understands the "require" attribute defined in (section 3.4.4) Example: Require:org.ietf.sip.subscribe,org.ietf.sdp.require A client SHOULD only include a Require: header where it truly requires the server to reject the request if the option is not supported.3.5.5. PINT URLs within PINT requests
Normally the hostnames and domain names that appear in the PINT URLs are the internal affair of each individual PINT system. A client uses the appropriate SDP payload to indicate the particular service it wishes to invoke; it is not necessary to use a particular URL to identify the service. A PINT URL is used in two different ways within PINT requests: within the Request-URI, and within the To: and From: headers. Use within the Request-URI requires clarification in order to ensure smooth interworking with the Telephone Network serviced by the PINT infrastructure, and this is covered next.
3.5.5.1. PINT URLS within Request-URIs
There are some occasions when it may be useful to indicate service information within the URL in a standardized way: a. it may not be possible to use SDP information to route the request if it is encrypted; b. it allows implementation that make use of I.N. "service indicators"; c. It enables multiple competing PINT gateways to REGISTER with a single "broker" server (proxy or redirect) (see section 6.3) For these reasons, the following conventions for URLs are offered for use in PINT requests: 1. The user portion of a sip URL indicates the service to be requested. At present the following services are defined: R2C (for Request-to-Call) R2F (for Request-to-Fax) R2HC (for Request-to-Hear-Content) The user portions "R2C", "R2F", and "R2HC" are reserved for the PINT milestone services. Other user portions MUST be used in case the requested service is not one of the Milestone services. See section 6.2 for some related considerations concerning registrations by competing PINT systems to a single PINT proxy server acting as a service broker. 2. The host portion of a sip URL contains the domain name of the PINT service provider. 3. A new url-parameter is defined to be "tsp" (for "telephone service provider"). This can be used to indicate the actual telephone network provider to be used to fulfill the PINT request. Thus, for example:- INVITE sip:R2C@pint.pintservice.com SIP/2.0 INVITE sip:R2F@pint.pintservice.com;tsp=telco.com SIP/2.0 INVITE sip:R2HC@pint.mycom.com;tsp=pbx23.mycom.com SIP/2.0 INVITE sip:13@pint.telco.com SIP/2.03.5.6. Telephony Network Parameters within PINT URLs
Any legal SIP URL can appear as a PINT URL within the Request-URI or To: header of a PINT request. But if the address is a telephone address, we indicated in section 3.4.3 that it may be necessary to include more information in order correctly to identify the remote
telephone terminal or service. PINT clients MAY include these attribute tags within PINT URLs if they are necessary or a useful complement to the telephone number within the SIP URL. These attribute tags MUST be included as URL parameters as defined in [1] (i.e. in the semi-colon separated manner). The following is an example of a PINT URL containing extra attribute tags: sip:+9725228808@pint.br.com;user=phone;require=Q763-plan;a=Q763-plan:4 As we noted in section 3.4.3, these extra attribute parameters will not normally be needed within a URL, because there is a great deal of context available to help the server interpret the phone number correctly. In particular, there is the SIP URL within the To: header, and there is also the Request-URI. In most cases this provides sufficient information for the telephone network. The SDP attributes defined in section 3 above will normally only be used when they are needed to supply necessary context to identify a telephone terminal.3.5.7. REGISTER requests within PINT
A PINT gateway is a SIP user agent server. A User Agent Server uses the REGISTER request to tell a proxy or redirect server that it is available to "receive calls" (i.e. to service requests). Thus a PINT Gateway registers with a proxy or redirect server the service that is accessible via itself, whilst in SIP, a user is registering his/her presence at a particular SIP Server. There may be competing PINT servers that can offer the same PINT service trying to register at a single PINT server. The PINT server might act as a "broker" among the various PINT gateways that can fulfill a request. A format for PINT URLs was specified in section 3.5.5 that enables independent PINT systems to REGISTER an offer to provide the same service. The registrar can apply its own mechanisms and policies to decide how to respond to INVITEs from clients seeking service (See section 6.3 for some possible deployment options). There is no change between SIP and PINT REGISTER semantics or syntax. Of course, the information in the PINT URLs within the REGISTER request may not be sufficient to completely define the service that a gateway can offer. The use of SIP and SDP within PINT REGISTER requests to enable a gateway to specify in more detail the services it can offer is the subject of future study.
3.5.8. BYE Requests in PINT
The semantics of BYE requests within PINT requires some extra precision. One issue concerns conferences that "cannot be left", and the other concerns keeping call state after the BYE. The BYE request [1] is normally used to indicate that the originating entity no longer wishes to be involved in the specified call. The request terminates the call and the media session. Applying this model to PINT, if a PINT client makes a request that results in invocation of a telephone call from A to B, a BYE request from the client, if accepted, should result in a termination of the phone call. One might expect this to be the case if the telephone call has not started when the BYE request is received. For example, if a request to fax is sent with a t= line indicating that the fax is to be sent tomorrow at 4 AM, the requestor might wish to cancel the request before the specified time. However, even if the call has yet to start, it may not be possible to terminate the media session on the telephone system side. For example, the fax call may be in progress when the BYE arrives, and perhaps it is just not possible to cancel the fax in session. Another possibility is that the entire telephone-side service might be completed before the BYE is received. In the above Request-to-Fax example, the BYE might be sent the following morning, and the entire fax has been sent before the BYE was received. It is too late to send the BYE. In the case where the telephone network cannot terminate the call, the server MUST return a "606 Not Acceptable" response to the BYE, along with a session description that indicates the telephone network session that is causing the problem. Thus, in PINT, a "Not Acceptable" response MAY be returned both to INVITE and BYE requests. It indicates that some aspect of the session description makes the request unacceptable. By allowing a server to return a "Not Acceptable" response to BYE requests, we are not changing its semantics, just enlarging its use. A combination of Warning: headers and i= lines within the session description can be used to indicate the precise nature of the problem.
Example:
SIP/2.0 606 Not Acceptable
From: ...
To: .......
.....
Warning: 399 pint.mycom.com Fax in progress, service cannot be
aborted
Content-Type: application/sdp
Content-Length: ...
v=0
...
...
i=3 of 5 pages sent OK
c=TN RFC2543 +12014064090
m=image 1 fax tif
a=fmtp:tif uri:http://tifsRus.com/yyyyyy.tif
Note that the server might return an updated session description
within a successful response to a BYE as well. This can be used, for
example, to indicate the actual start times and stop times of the
telephone session, or how many pages were sent in the fax
transmission.
The second issue concerns how long must a server keep call state
after receiving a BYE. A question arises because other clients might
still wish to send queries about the telephone network session that
was the subject of the PINT transaction. Ordinary SIP semantics have
three important implications for this situation:
1. A BYE indicates that the requesting client will clear out all call
state as soon as it receives a successful response. A client SHOULD
NOT send a SUBSCRIBE request after it has sent a BYE.
2. A server may return an Expires: header within a successful
response to a BYE request. This indicates for how long the server
will retain session state about the telephone network session. At any
point during this time, a client may send a SUBSCRIBE request to the
server to learn about the session state (although as explained in the
previous paragraph, a client that has sent a BYE will not normally
send a SUBSCRIBE).
3. When engaged in a SUBSCRIBE/NOTIFY monitoring session, PINT
servers that send UNSUBSCRIBE to a URL listed in the Contact: header
of a client request SHOULD not clear session state until after the
successful response to the UNSUBSCRIBE message is received. For
example, it may be that the requesting client host is turned off (or
in a low power mode) when the telephone service is executed (and is therefore not available at the location previously specified in the Contact: attribute) to receive the PINT server's UNSUBSCRIBE. Of course, it is possible that the UNSUBSCRIBE request will simply time out.
(page 37 continued on part 3)