7
Network Working Group M. Handley
8
Request for Comments: 2327 V. Jacobson
9
Category: Standards Track ISI/LBNL
13
SDP: Session Description Protocol
17
This document specifies an Internet standards track protocol for the
18
Internet community, and requests discussion and suggestions for
19
improvements. Please refer to the current edition of the "Internet
20
Official Protocol Standards" (STD 1) for the standardization state
21
and status of this protocol. Distribution of this memo is unlimited.
25
Copyright (C) The Internet Society (1998). All Rights Reserved.
29
This document defines the Session Description Protocol, SDP. SDP is
30
intended for describing multimedia sessions for the purposes of
31
session announcement, session invitation, and other forms of
32
multimedia session initiation.
34
This document is a product of the Multiparty Multimedia Session
35
Control (MMUSIC) working group of the Internet Engineering Task
36
Force. Comments are solicited and should be addressed to the working
37
group's mailing list at confctrl@isi.edu and/or the authors.
41
On the Internet multicast backbone (Mbone), a session directory tool
42
is used to advertise multimedia conferences and communicate the
43
conference addresses and conference tool-specific information
44
necessary for participation. This document defines a session
45
description protocol for this purpose, and for general real-time
46
multimedia session description purposes. This memo does not describe
47
multicast address allocation or the distribution of SDP messages in
48
detail. These are described in accompanying memos. SDP is not
49
intended for negotiation of media encodings.
58
Handley & Jacobson Standards Track [Page 1]
60
RFC 2327 SDP April 1998
65
The Mbone is the part of the internet that supports IP multicast, and
66
thus permits efficient many-to-many communication. It is used
67
extensively for multimedia conferencing. Such conferences usually
68
have the property that tight coordination of conference membership is
69
not necessary; to receive a conference, a user at an Mbone site only
70
has to know the conference's multicast group address and the UDP
71
ports for the conference data streams.
73
Session directories assist the advertisement of conference sessions
74
and communicate the relevant conference setup information to
75
prospective participants. SDP is designed to convey such information
76
to recipients. SDP is purely a format for session description - it
77
does not incorporate a transport protocol, and is intended to use
78
different transport protocols as appropriate including the Session
79
Announcement Protocol [4], Session Initiation Protocol [11], Real-
80
Time Streaming Protocol [12], electronic mail using the MIME
81
extensions, and the Hypertext Transport Protocol.
83
SDP is intended to be general purpose so that it can be used for a
84
wider range of network environments and applications than just
85
multicast session directories. However, it is not intended to
86
support negotiation of session content or media encodings - this is
87
viewed as outside the scope of session description.
91
The following terms are used in this document, and have specific
92
meaning within the context of this document.
95
A multimedia conference is a set of two or more communicating users
96
along with the software they are using to communicate.
99
A multimedia session is a set of multimedia senders and receivers
100
and the data streams flowing from senders to receivers. A
101
multimedia conference is an example of a multimedia session.
103
Session Advertisement
104
See session announcement.
107
A session announcement is a mechanism by which a session
108
description is conveyed to users in a proactive fashion, i.e., the
109
session description was not explicitly requested by the user.
114
Handley & Jacobson Standards Track [Page 2]
116
RFC 2327 SDP April 1998
120
A well defined format for conveying sufficient information to
121
discover and participate in a multimedia session.
125
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
126
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
127
document are to be interpreted as described in RFC 2119.
131
4.1. Multicast Announcements
133
SDP is a session description protocol for multimedia sessions. A
134
common mode of usage is for a client to announce a conference session
135
by periodically multicasting an announcement packet to a well known
136
multicast address and port using the Session Announcement Protocol
139
SAP packets are UDP packets with the following format:
141
|--------------------|
143
|--------------------|
148
The header is the Session Announcement Protocol header. SAP is
149
described in more detail in a companion memo [4]
151
The text payload is an SDP session description, as described in this
152
memo. The text payload should be no greater than 1 Kbyte in length.
153
If announced by SAP, only one session announcement is permitted in a
156
4.2. Email and WWW Announcements
158
Alternative means of conveying session descriptions include
159
electronic mail and the World Wide Web. For both email and WWW
160
distribution, the use of the MIME content type "application/sdp"
161
should be used. This enables the automatic launching of applications
162
for participation in the session from the WWW client or mail reader
163
in a standard manner.
170
Handley & Jacobson Standards Track [Page 3]
172
RFC 2327 SDP April 1998
175
Note that announcements of multicast sessions made only via email or
176
the World Wide Web (WWW) do not have the property that the receiver
177
of a session announcement can necessarily receive the session because
178
the multicast sessions may be restricted in scope, and access to the
179
WWW server or reception of email is possible outside this scope. SAP
180
announcements do not suffer from this mismatch.
182
5. Requirements and Recommendations
184
The purpose of SDP is to convey information about media streams in
185
multimedia sessions to allow the recipients of a session description
186
to participate in the session. SDP is primarily intended for use in
187
an internetwork, although it is sufficiently general that it can
188
describe conferences in other network environments.
190
A multimedia session, for these purposes, is defined as a set of
191
media streams that exist for some duration of time. Media streams
192
can be many-to-many. The times during which the session is active
193
need not be continuous.
195
Thus far, multicast based sessions on the Internet have differed from
196
many other forms of conferencing in that anyone receiving the traffic
197
can join the session (unless the session traffic is encrypted). In
198
such an environment, SDP serves two primary purposes. It is a means
199
to communicate the existence of a session, and is a means to convey
200
sufficient information to enable joining and participating in the
201
session. In a unicast environment, only the latter purpose is likely
206
o Session name and purpose
208
o Time(s) the session is active
210
o The media comprising the session
212
o Information to receive those media (addresses, ports, formats and
215
As resources necessary to participate in a session may be limited,
216
some additional information may also be desirable:
218
o Information about the bandwidth to be used by the conference
220
o Contact information for the person responsible for the session
226
Handley & Jacobson Standards Track [Page 4]
228
RFC 2327 SDP April 1998
231
In general, SDP must convey sufficient information to be able to join
232
a session (with the possible exception of encryption keys) and to
233
announce the resources to be used to non-participants that may need
236
5.1. Media Information
240
o The type of media (video, audio, etc)
242
o The transport protocol (RTP/UDP/IP, H.320, etc)
244
o The format of the media (H.261 video, MPEG video, etc)
246
For an IP multicast session, the following are also conveyed:
248
o Multicast address for media
250
o Transport Port for media
252
This address and port are the destination address and destination
253
port of the multicast stream, whether being sent, received, or both.
255
For an IP unicast session, the following are conveyed:
257
o Remote address for media
259
o Transport port for contact address
261
The semantics of this address and port depend on the media and
262
transport protocol defined. By default, this is the remote address
263
and remote port to which data is sent, and the remote address and
264
local port on which to receive data. However, some media may define
265
to use these to establish a control channel for the actual media
268
5.2. Timing Information
270
Sessions may either be bounded or unbounded in time. Whether or not
271
they are bounded, they may be only active at specific times.
275
o An arbitrary list of start and stop times bounding the session
277
o For each bound, repeat times such as "every Wednesday at 10am for
282
Handley & Jacobson Standards Track [Page 5]
284
RFC 2327 SDP April 1998
287
This timing information is globally consistent, irrespective of local
288
time zone or daylight saving time.
290
5.3. Private Sessions
292
It is possible to create both public sessions and private sessions.
293
Private sessions will typically be conveyed by encrypting the session
294
description to distribute it. The details of how encryption is
295
performed are dependent on the mechanism used to convey SDP - see [4]
296
for how this is done for session announcements.
298
If a session announcement is private it is possible to use that
299
private announcement to convey encryption keys necessary to decode
300
each of the media in a conference, including enough information to
301
know which encryption scheme is used for each media.
303
5.4. Obtaining Further Information about a Session
305
A session description should convey enough information to decide
306
whether or not to participate in a session. SDP may include
307
additional pointers in the form of Universal Resources Identifiers
308
(URIs) for more information about the session.
312
When many session descriptions are being distributed by SAP or any
313
other advertisement mechanism, it may be desirable to filter
314
announcements that are of interest from those that are not. SDP
315
supports a categorisation mechanism for sessions that is capable of
318
5.6. Internationalization
320
The SDP specification recommends the use of the ISO 10646 character
321
sets in the UTF-8 encoding (RFC 2044) to allow many different
322
languages to be represented. However, to assist in compact
323
representations, SDP also allows other character sets such as ISO
324
8859-1 to be used when desired. Internationalization only applies to
325
free-text fields (session name and background information), and not
330
SDP session descriptions are entirely textual using the ISO 10646
331
character set in UTF-8 encoding. SDP field names and attributes names
332
use only the US-ASCII subset of UTF-8, but textual fields and
333
attribute values may use the full ISO 10646 character set. The
334
textual form, as opposed to a binary encoding such as ASN/1 or XDR,
338
Handley & Jacobson Standards Track [Page 6]
340
RFC 2327 SDP April 1998
343
was chosen to enhance portability, to enable a variety of transports
344
to be used (e.g, session description in a MIME email message) and to
345
allow flexible, text-based toolkits (e.g., Tcl/Tk ) to be used to
346
generate and to process session descriptions. However, since the
347
total bandwidth allocated to all SAP announcements is strictly
348
limited, the encoding is deliberately compact. Also, since
349
announcements may be transported via very unreliable means (e.g.,
350
email) or damaged by an intermediate caching server, the encoding was
351
designed with strict order and formatting rules so that most errors
352
would result in malformed announcements which could be detected
353
easily and discarded. This also allows rapid discarding of encrypted
354
announcements for which a receiver does not have the correct key.
356
An SDP session description consists of a number of lines of text of
357
the form <type>=<value> <type> is always exactly one character and is
358
case-significant. <value> is a structured text string whose format
359
depends on <type>. It also will be case-significant unless a
360
specific field defines otherwise. Whitespace is not permitted either
361
side of the `=' sign. In general <value> is either a number of fields
362
delimited by a single space character or a free format string.
364
A session description consists of a session-level description
365
(details that apply to the whole session and all media streams) and
366
optionally several media-level descriptions (details that apply onto
367
to a single media stream).
369
An announcement consists of a session-level section followed by zero
370
or more media-level sections. The session-level part starts with a
371
`v=' line and continues to the first media-level section. The media
372
description starts with an `m=' line and continues to the next media
373
description or end of the whole session description. In general,
374
session-level values are the default for all media unless overridden
375
by an equivalent media-level value.
377
When SDP is conveyed by SAP, only one session description is allowed
378
per packet. When SDP is conveyed by other means, many SDP session
379
descriptions may be concatenated together (the `v=' line indicating
380
the start of a session description terminates the previous
381
description). Some lines in each description are required and some
382
are optional but all must appear in exactly the order given here (the
383
fixed order greatly enhances error detection and allows for a simple
384
parser). Optional items are marked with a `*'.
387
v= (protocol version)
388
o= (owner/creator and session identifier).
390
i=* (session information)
394
Handley & Jacobson Standards Track [Page 7]
396
RFC 2327 SDP April 1998
399
u=* (URI of description)
402
c=* (connection information - not required if included in all media)
403
b=* (bandwidth information)
404
One or more time descriptions (see below)
405
z=* (time zone adjustments)
407
a=* (zero or more session attribute lines)
408
Zero or more media descriptions (see below)
411
t= (time the session is active)
412
r=* (zero or more repeat times)
415
m= (media name and transport address)
417
c=* (connection information - optional if included at session-level)
418
b=* (bandwidth information)
420
a=* (zero or more media attribute lines)
422
The set of `type' letters is deliberately small and not intended to
423
be extensible -- SDP parsers must completely ignore any announcement
424
that contains a `type' letter that it does not understand. The
425
`attribute' mechanism ("a=" described below) is the primary means for
426
extending SDP and tailoring it to particular applications or media.
427
Some attributes (the ones listed in this document) have a defined
428
meaning but others may be added on an application-, media- or
429
session-specific basis. A session directory must ignore any
430
attribute it doesn't understand.
432
The connection (`c=') and attribute (`a=') information in the
433
session-level section applies to all the media of that session unless
434
overridden by connection information or an attribute of the same name
435
in the media description. For instance, in the example below, each
436
media behaves as if it were given a `recvonly' attribute.
438
An example SDP description is:
441
o=mhandley 2890844526 2890842807 IN IP4 126.16.64.4
443
i=A Seminar on the session description protocol
444
u=http://www.cs.ucl.ac.uk/staff/M.Handley/sdp.03.ps
445
e=mjh@isi.edu (Mark Handley)
446
c=IN IP4 224.2.17.12/127
450
Handley & Jacobson Standards Track [Page 8]
452
RFC 2327 SDP April 1998
455
t=2873397496 2873404696
457
m=audio 49170 RTP/AVP 0
458
m=video 51372 RTP/AVP 31
459
m=application 32416 udp wb
462
Text records such as the session name and information are bytes
463
strings which may contain any byte with the exceptions of 0x00 (Nul),
464
0x0a (ASCII newline) and 0x0d (ASCII carriage return). The sequence
465
CRLF (0x0d0a) is used to end a record, although parsers should be
466
tolerant and also accept records terminated with a single newline
467
character. By default these byte strings contain ISO-10646
468
characters in UTF-8 encoding, but this default may be changed using
469
the `charset' attribute.
475
The "v=" field gives the version of the Session Description Protocol.
476
There is no minor version number.
480
o=<username> <session id> <version> <network type> <address type>
483
The "o=" field gives the originator of the session (their username
484
and the address of the user's host) plus a session id and session
487
<username> is the user's login on the originating host, or it is "-"
488
if the originating host does not support the concept of user ids.
489
<username> must not contain spaces. <session id> is a numeric string
490
such that the tuple of <username>, <session id>, <network type>,
491
<address type> and <address> form a globally unique identifier for
494
The method of <session id> allocation is up to the creating tool, but
495
it has been suggested that a Network Time Protocol (NTP) timestamp be
496
used to ensure uniqueness [1].
498
<version> is a version number for this announcement. It is needed
499
for proxy announcements to detect which of several announcements for
500
the same session is the most recent. Again its usage is up to the
506
Handley & Jacobson Standards Track [Page 9]
508
RFC 2327 SDP April 1998
511
creating tool, so long as <version> is increased when a modification
512
is made to the session data. Again, it is recommended (but not
513
mandatory) that an NTP timestamp is used.
515
<network type> is a text string giving the type of network.
516
Initially "IN" is defined to have the meaning "Internet". <address
517
type> is a text string giving the type of the address that follows.
518
Initially "IP4" and "IP6" are defined. <address> is the globally
519
unique address of the machine from which the session was created.
520
For an address type of IP4, this is either the fully-qualified domain
521
name of the machine, or the dotted-decimal representation of the IP
522
version 4 address of the machine. For an address type of IP6, this
523
is either the fully-qualified domain name of the machine, or the
524
compressed textual representation of the IP version 6 address of the
525
machine. For both IP4 and IP6, the fully-qualified domain name is
526
the form that SHOULD be given unless this is unavailable, in which
527
case the globally unique address may be substituted. A local IP
528
address MUST NOT be used in any context where the SDP description
529
might leave the scope in which the address is meaningful.
531
In general, the "o=" field serves as a globally unique identifier for
532
this version of this session description, and the subfields excepting
533
the version taken together identify the session irrespective of any
540
The "s=" field is the session name. There must be one and only one
541
"s=" field per session description, and it must contain ISO 10646
542
characters (but see also the `charset' attribute below).
544
Session and Media Information
546
i=<session description>
548
The "i=" field is information about the session. There may be at
549
most one session-level "i=" field per session description, and at
550
most one "i=" field per media. Although it may be omitted, this is
551
discouraged for session announcements, and user interfaces for
552
composing sessions should require text to be entered. If it is
553
present it must contain ISO 10646 characters (but see also the
554
`charset' attribute below).
556
A single "i=" field can also be used for each media definition. In
557
media definitions, "i=" fields are primarily intended for labeling
558
media streams. As such, they are most likely to be useful when a
562
Handley & Jacobson Standards Track [Page 10]
564
RFC 2327 SDP April 1998
567
single session has more than one distinct media stream of the same
568
media type. An example would be two different whiteboards, one for
569
slides and one for feedback and questions.
575
o A URI is a Universal Resource Identifier as used by WWW clients
577
o The URI should be a pointer to additional information about the
580
o This field is optional, but if it is present it should be specified
581
before the first media field
583
o No more than one URI field is allowed per session description
586
Email Address and Phone Number
591
o These specify contact information for the person responsible for
592
the conference. This is not necessarily the same person that
593
created the conference announcement.
595
o Either an email field or a phone field must be specified.
596
Additional email and phone fields are allowed.
598
o If these are present, they should be specified before the first
601
o More than one email or phone field can be given for a session
604
o Phone numbers should be given in the conventional international
606
format - preceded by a "+ and the international country code.
607
There must be a space or a hyphen ("-") between the country code
608
and the rest of the phone number. Spaces and hyphens may be used
609
to split up a phone field to aid readability if desired. For
612
p=+44-171-380-7777 or p=+1 617 253 6011
618
Handley & Jacobson Standards Track [Page 11]
620
RFC 2327 SDP April 1998
623
o Both email addresses and phone numbers can have an optional free
624
text string associated with them, normally giving the name of the
625
person who may be contacted. This should be enclosed in
626
parenthesis if it is present. For example:
628
e=mjh@isi.edu (Mark Handley)
630
The alternative RFC822 name quoting convention is also allowed for
631
both email addresses and phone numbers. For example,
633
e=Mark Handley <mjh@isi.edu>
635
The free text string should be in the ISO-10646 character set with
636
UTF-8 encoding, or alternatively in ISO-8859-1 or other encodings
637
if the appropriate charset session-level attribute is set.
641
c=<network type> <address type> <connection address>
643
The "c=" field contains connection data.
645
A session announcement must contain one "c=" field in each media
646
description (see below) or a "c=" field at the session-level. It may
647
contain a session-level "c=" field and one additional "c=" field per
648
media description, in which case the per-media values override the
649
session-level settings for the relevant media.
651
The first sub-field is the network type, which is a text string
652
giving the type of network. Initially "IN" is defined to have the
655
The second sub-field is the address type. This allows SDP to be used
656
for sessions that are not IP based. Currently only IP4 is defined.
658
The third sub-field is the connection address. Optional extra
659
subfields may be added after the connection address depending on the
660
value of the <address type> field.
662
For IP4 addresses, the connection address is defined as follows:
664
o Typically the connection address will be a class-D IP multicast
666
group address. If the session is not multicast, then the
667
connection address contains the fully-qualified domain name or the
668
unicast IP address of the expected data source or data relay or
669
data sink as determined by additional attribute fields. It is not
670
expected that fully-qualified domain names or unicast addresses
674
Handley & Jacobson Standards Track [Page 12]
676
RFC 2327 SDP April 1998
679
will be given in a session description that is communicated by a
680
multicast announcement, though this is not prohibited. If a
681
unicast data stream is to pass through a network address
682
translator, the use of a fully-qualified domain name rather than an
683
unicast IP address is RECOMMENDED. In other cases, the use of an
684
IP address to specify a particular interface on a multi-homed host
685
might be required. Thus this specification leaves the decision as
686
to which to use up to the individual application, but all
687
applications MUST be able to cope with receiving both formats.
689
o Conferences using an IP multicast connection address must also have
690
a time to live (TTL) value present in addition to the multicast
691
address. The TTL and the address together define the scope with
692
which multicast packets sent in this conference will be sent. TTL
693
values must be in the range 0-255.
695
The TTL for the session is appended to the address using a slash as
696
a separator. An example is:
698
c=IN IP4 224.2.1.1/127
700
Hierarchical or layered encoding schemes are data streams where the
701
encoding from a single media source is split into a number of
702
layers. The receiver can choose the desired quality (and hence
703
bandwidth) by only subscribing to a subset of these layers. Such
704
layered encodings are normally transmitted in multiple multicast
705
groups to allow multicast pruning. This technique keeps unwanted
706
traffic from sites only requiring certain levels of the hierarchy.
707
For applications requiring multiple multicast groups, we allow the
708
following notation to be used for the connection address:
710
<base multicast address>/<ttl>/<number of addresses>
712
If the number of addresses is not given it is assumed to be one.
713
Multicast addresses so assigned are contiguously allocated above
714
the base address, so that, for example:
716
c=IN IP4 224.2.1.1/127/3
718
would state that addresses 224.2.1.1, 224.2.1.2 and 224.2.1.3 are
719
to be used at a ttl of 127. This is semantically identical to
720
including multiple "c=" lines in a media description:
722
c=IN IP4 224.2.1.1/127
723
c=IN IP4 224.2.1.2/127
724
c=IN IP4 224.2.1.3/127
730
Handley & Jacobson Standards Track [Page 13]
732
RFC 2327 SDP April 1998
735
Multiple addresses or "c=" lines can only be specified on a per-
736
media basis, and not for a session-level "c=" field.
738
It is illegal for the slash notation described above to be used for
739
IP unicast addresses.
743
b=<modifier>:<bandwidth-value>
745
o This specifies the proposed bandwidth to be used by the session or
746
media, and is optional.
748
o <bandwidth-value> is in kilobits per second
750
o <modifier> is a single alphanumeric word giving the meaning of the
753
o Two modifiers are initially defined:
755
CT Conference Total: An implicit maximum bandwidth is associated with
756
each TTL on the Mbone or within a particular multicast
757
administrative scope region (the Mbone bandwidth vs. TTL limits are
758
given in the MBone FAQ). If the bandwidth of a session or media in
759
a session is different from the bandwidth implicit from the scope,
760
a `b=CT:...' line should be supplied for the session giving the
761
proposed upper limit to the bandwidth used. The primary purpose of
762
this is to give an approximate idea as to whether two or more
763
conferences can co-exist simultaneously.
765
AS Application-Specific Maximum: The bandwidth is interpreted to be
766
application-specific, i.e., will be the application's concept of
767
maximum bandwidth. Normally this will coincide with what is set on
768
the application's "maximum bandwidth" control if applicable.
770
Note that CT gives a total bandwidth figure for all the media at
771
all sites. AS gives a bandwidth figure for a single media at a
772
single site, although there may be many sites sending
775
o Extension Mechanism: Tool writers can define experimental bandwidth
776
modifiers by prefixing their modifier with "X-". For example:
780
SDP parsers should ignore bandwidth fields with unknown modifiers.
781
Modifiers should be alpha-numeric and, although no length limit is
782
given, they are recommended to be short.
786
Handley & Jacobson Standards Track [Page 14]
788
RFC 2327 SDP April 1998
791
Times, Repeat Times and Time Zones
793
t=<start time> <stop time>
795
o "t=" fields specify the start and stop times for a conference
796
session. Multiple "t=" fields may be used if a session is active
797
at multiple irregularly spaced times; each additional "t=" field
798
specifies an additional period of time for which the session will
799
be active. If the session is active at regular times, an "r="
800
field (see below) should be used in addition to and following a
801
"t=" field - in which case the "t=" field specifies the start and
802
stop times of the repeat sequence.
804
o The first and second sub-fields give the start and stop times for
805
the conference respectively. These values are the decimal
806
representation of Network Time Protocol (NTP) time values in
807
seconds [1]. To convert these values to UNIX time, subtract
810
o If the stop-time is set to zero, then the session is not bounded,
811
though it will not become active until after the start-time. If
812
the start-time is also zero, the session is regarded as permanent.
814
User interfaces should strongly discourage the creation of
815
unbounded and permanent sessions as they give no information about
816
when the session is actually going to terminate, and so make
817
scheduling difficult.
819
The general assumption may be made, when displaying unbounded
820
sessions that have not timed out to the user, that an unbounded
821
session will only be active until half an hour from the current
822
time or the session start time, whichever is the later. If
823
behaviour other than this is required, an end-time should be given
824
and modified as appropriate when new information becomes available
825
about when the session should really end.
827
Permanent sessions may be shown to the user as never being active
828
unless there are associated repeat times which state precisely when
829
the session will be active. In general, permanent sessions should
830
not be created for any session expected to have a duration of less
831
than 2 months, and should be discouraged for sessions expected to
832
have a duration of less than 6 months.
834
r=<repeat interval> <active duration> <list of offsets from start-
837
o "r=" fields specify repeat times for a session. For example, if
838
a session is active at 10am on Monday and 11am on Tuesday for one
842
Handley & Jacobson Standards Track [Page 15]
844
RFC 2327 SDP April 1998
847
hour each week for three months, then the <start time> in the
848
corresponding "t=" field would be the NTP representation of 10am on
849
the first Monday, the <repeat interval> would be 1 week, the
850
<active duration> would be 1 hour, and the offsets would be zero
851
and 25 hours. The corresponding "t=" field stop time would be the
852
NTP representation of the end of the last session three months
853
later. By default all fields are in seconds, so the "r=" and "t="
856
t=3034423619 3042462419
857
r=604800 3600 0 90000
859
To make announcements more compact, times may also be given in units
860
of days, hours or minutes. The syntax for these is a number
861
immediately followed by a single case-sensitive character.
862
Fractional units are not allowed - a smaller unit should be used
863
instead. The following unit specification characters are allowed:
865
d - days (86400 seconds)
866
h - minutes (3600 seconds)
867
m - minutes (60 seconds)
868
s - seconds (allowed for completeness but not recommended)
870
Thus, the above announcement could also have been written:
874
Monthly and yearly repeats cannot currently be directly specified
875
with a single SDP repeat time - instead separate "t" fields should
876
be used to explicitly list the session times.
878
z=<adjustment time> <offset> <adjustment time> <offset> ....
880
o To schedule a repeated session which spans a change from daylight-
881
saving time to standard time or vice-versa, it is necessary to
882
specify offsets from the base repeat times. This is required
883
because different time zones change time at different times of day,
884
different countries change to or from daylight time on different
885
dates, and some countries do not have daylight saving time at all.
887
Thus in order to schedule a session that is at the same time winter
888
and summer, it must be possible to specify unambiguously by whose
889
time zone a session is scheduled. To simplify this task for
890
receivers, we allow the sender to specify the NTP time that a time
891
zone adjustment happens and the offset from the time when the
892
session was first scheduled. The "z" field allows the sender to
893
specify a list of these adjustment times and offsets from the base
898
Handley & Jacobson Standards Track [Page 16]
900
RFC 2327 SDP April 1998
905
z=2882844526 -1h 2898848070 0
907
This specifies that at time 2882844526 the time base by which the
908
session's repeat times are calculated is shifted back by 1 hour,
909
and that at time 2898848070 the session's original time base is
910
restored. Adjustments are always relative to the specified start
911
time - they are not cumulative.
913
o If a session is likely to last several years, it is expected
915
the session announcement will be modified periodically rather than
916
transmit several years worth of adjustments in one announcement.
921
k=<method>:<encryption key>
923
o The session description protocol may be used to convey encryption
924
keys. A key field is permitted before the first media entry (in
925
which case it applies to all media in the session), or for each
926
media entry as required.
928
o The format of keys and their usage is outside the scope of this
929
document, but see [3].
931
o The method indicates the mechanism to be used to obtain a usable
932
key by external means, or from the encoded encryption key given.
934
The following methods are defined:
936
k=clear:<encryption key>
937
The encryption key (as described in [3] for RTP media streams
938
under the AV profile) is included untransformed in this key
941
k=base64:<encoded encryption key>
942
The encryption key (as described in [3] for RTP media streams
943
under the AV profile) is included in this key field but has been
944
base64 encoded because it includes characters that are
947
k=uri:<URI to obtain key>
948
A Universal Resource Identifier as used by WWW clients is
949
included in this key field. The URI refers to the data
950
containing the key, and may require additional authentication
954
Handley & Jacobson Standards Track [Page 17]
956
RFC 2327 SDP April 1998
959
before the key can be returned. When a request is made to the
960
given URI, the MIME content-type of the reply specifies the
961
encoding for the key in the reply. The key should not be
962
obtained until the user wishes to join the session to reduce
963
synchronisation of requests to the WWW server(s).
966
No key is included in this SDP description, but the session or
967
media stream referred to by this key field is encrypted. The
968
user should be prompted for the key when attempting to join the
969
session, and this user-supplied key should then be used to
970
decrypt the media streams.
975
a=<attribute>:<value>
977
Attributes are the primary means for extending SDP. Attributes may
978
be defined to be used as "session-level" attributes, "media-level"
981
A media description may have any number of attributes ("a=" fields)
982
which are media specific. These are referred to as "media-level"
983
attributes and add information about the media stream. Attribute
984
fields can also be added before the first media field; these
985
"session-level" attributes convey additional information that applies
986
to the conference as a whole rather than to individual media; an
987
example might be the conference's floor control policy.
989
Attribute fields may be of two forms:
991
o property attributes. A property attribute is simply of the form
992
"a=<flag>". These are binary attributes, and the presence of the
993
attribute conveys that the attribute is a property of the session.
994
An example might be "a=recvonly".
996
o value attributes. A value attribute is of the form
997
"a=<attribute>:<value>". An example might be that a whiteboard
998
could have the value attribute "a=orient:landscape"
1000
Attribute interpretation depends on the media tool being invoked.
1001
Thus receivers of session descriptions should be configurable in
1002
their interpretation of announcements in general and of attributes in
1005
Attribute names must be in the US-ASCII subset of ISO-10646/UTF-8.
1010
Handley & Jacobson Standards Track [Page 18]
1012
RFC 2327 SDP April 1998
1015
Attribute values are byte strings, and MAY use any byte value except
1016
0x00 (Nul), 0x0A (LF), and 0x0D (CR). By default, attribute values
1017
are to be interpreted as in ISO-10646 character set with UTF-8
1018
encoding. Unlike other text fields, attribute values are NOT
1019
normally affected by the `charset' attribute as this would make
1020
comparisons against known values problematic. However, when an
1021
attribute is defined, it can be defined to be charset-dependent, in
1022
which case it's value should be interpreted in the session charset
1023
rather than in ISO-10646.
1025
Attributes that will be commonly used can be registered with IANA
1026
(see Appendix B). Unregistered attributes should begin with "X-" to
1027
prevent inadvertent collision with registered attributes. In either
1028
case, if an attribute is received that is not understood, it should
1029
simply be ignored by the receiver.
1033
m=<media> <port> <transport> <fmt list>
1035
A session description may contain a number of media descriptions.
1036
Each media description starts with an "m=" field, and is terminated
1037
by either the next "m=" field or by the end of the session
1038
description. A media field also has several sub-fields:
1040
o The first sub-field is the media type. Currently defined media are
1041
"audio", "video", "application", "data" and "control", though this
1042
list may be extended as new communication modalities emerge (e.g.,
1043
telepresense). The difference between "application" and "data" is
1044
that the former is a media flow such as whiteboard information, and
1045
the latter is bulk-data transfer such as multicasting of program
1046
executables which will not typically be displayed to the user.
1047
"control" is used to specify an additional conference control
1048
channel for the session.
1050
o The second sub-field is the transport port to which the media
1051
stream will be sent. The meaning of the transport port depends on
1052
the network being used as specified in the relevant "c" field and
1053
on the transport protocol defined in the third sub-field. Other
1054
ports used by the media application (such as the RTCP port, see
1055
[2]) should be derived algorithmically from the base media port.
1057
Note: For transports based on UDP, the value should be in the range
1058
1024 to 65535 inclusive. For RTP compliance it should be an even
1066
Handley & Jacobson Standards Track [Page 19]
1068
RFC 2327 SDP April 1998
1071
For applications where hierarchically encoded streams are being
1072
sent to a unicast address, it may be necessary to specify multiple
1073
transport ports. This is done using a similar notation to that
1074
used for IP multicast addresses in the "c=" field:
1076
m=<media> <port>/<number of ports> <transport> <fmt list>
1078
In such a case, the ports used depend on the transport protocol.
1079
For RTP, only the even ports are used for data and the
1080
corresponding one-higher odd port is used for RTCP. For example:
1082
m=video 49170/2 RTP/AVP 31
1084
would specify that ports 49170 and 49171 form one RTP/RTCP pair and
1085
49172 and 49173 form the second RTP/RTCP pair. RTP/AVP is the
1086
transport protocol and 31 is the format (see below).
1088
It is illegal for both multiple addresses to be specified in the
1089
"c=" field and for multiple ports to be specified in the "m=" field
1090
in the same session description.
1092
o The third sub-field is the transport protocol. The transport
1093
protocol values are dependent on the address-type field in the "c="
1094
fields. Thus a "c=" field of IP4 defines that the transport
1095
protocol runs over IP4. For IP4, it is normally expected that most
1096
media traffic will be carried as RTP over UDP. The following
1097
transport protocols are preliminarily defined, but may be extended
1098
through registration of new protocols with IANA:
1100
- RTP/AVP - the IETF's Realtime Transport Protocol using the
1101
Audio/Video profile carried over UDP.
1103
- udp - User Datagram Protocol
1105
If an application uses a single combined proprietary media format
1106
and transport protocol over UDP, then simply specifying the
1107
transport protocol as udp and using the format field to distinguish
1108
the combined protocol is recommended. If a transport protocol is
1109
used over UDP to carry several distinct media types that need to be
1110
distinguished by a session directory, then specifying the transport
1111
protocol and media format separately is necessary. RTP is an
1112
example of a transport-protocol that carries multiple payload
1113
formats that must be distinguished by the session directory for it
1114
to know how to start appropriate tools, relays, mixers or
1122
Handley & Jacobson Standards Track [Page 20]
1124
RFC 2327 SDP April 1998
1127
The main reason to specify the transport-protocol in addition to
1128
the media format is that the same standard media formats may be
1129
carried over different transport protocols even when the network
1130
protocol is the same - a historical example is vat PCM audio and
1131
RTP PCM audio. In addition, relays and monitoring tools that are
1132
transport-protocol-specific but format-independent are possible.
1134
For RTP media streams operating under the RTP Audio/Video Profile
1135
[3], the protocol field is "RTP/AVP". Should other RTP profiles be
1136
defined in the future, their profiles will be specified in the same
1137
way. For example, the protocol field "RTP/XYZ" would specify RTP
1138
operating under a profile whose short name is "XYZ".
1140
o The fourth and subsequent sub-fields are media formats. For audio
1141
and video, these will normally be a media payload type as defined
1142
in the RTP Audio/Video Profile.
1144
When a list of payload formats is given, this implies that all of
1145
these formats may be used in the session, but the first of these
1146
formats is the default format for the session.
1148
For media whose transport protocol is not RTP or UDP the format
1149
field is protocol specific. Such formats should be defined in an
1150
additional specification document.
1152
For media whose transport protocol is RTP, SDP can be used to
1153
provide a dynamic binding of media encoding to RTP payload type.
1154
The encoding names in the RTP AV Profile do not specify unique
1155
audio encodings (in terms of clock rate and number of audio
1156
channels), and so they are not used directly in SDP format fields.
1157
Instead, the payload type number should be used to specify the
1158
format for static payload types and the payload type number along
1159
with additional encoding information should be used for dynamically
1160
allocated payload types.
1162
An example of a static payload type is u-law PCM coded single
1163
channel audio sampled at 8KHz. This is completely defined in the
1164
RTP Audio/Video profile as payload type 0, so the media field for
1165
such a stream sent to UDP port 49232 is:
1167
m=video 49232 RTP/AVP 0
1169
An example of a dynamic payload type is 16 bit linear encoded
1170
stereo audio sampled at 16KHz. If we wish to use dynamic RTP/AVP
1171
payload type 98 for such a stream, additional information is
1172
required to decode it:
1174
m=video 49232 RTP/AVP 98
1178
Handley & Jacobson Standards Track [Page 21]
1180
RFC 2327 SDP April 1998
1183
a=rtpmap:98 L16/16000/2
1185
The general form of an rtpmap attribute is:
1187
a=rtpmap:<payload type> <encoding name>/<clock rate>[/<encoding
1190
For audio streams, <encoding parameters> may specify the number of
1191
audio channels. This parameter may be omitted if the number of
1192
channels is one provided no additional parameters are needed. For
1193
video streams, no encoding parameters are currently specified.
1195
Additional parameters may be defined in the future, but
1196
codecspecific parameters should not be added. Parameters added to
1197
an rtpmap attribute should only be those required for a session
1198
directory to make the choice of appropriate media too to
1199
participate in a session. Codec-specific parameters should be
1200
added in other attributes.
1202
Up to one rtpmap attribute can be defined for each media format
1203
specified. Thus we might have:
1205
m=audio 49230 RTP/AVP 96 97 98
1207
a=rtpmap:97 L16/8000
1208
a=rtpmap:98 L16/11025/2
1210
RTP profiles that specify the use of dynamic payload types must
1211
define the set of valid encoding names and/or a means to register
1212
encoding names if that profile is to be used with SDP.
1214
Experimental encoding formats can also be specified using rtpmap.
1215
RTP formats that are not registered as standard format names must
1216
be preceded by "X-". Thus a new experimental redundant audio
1217
stream called GSMLPC using dynamic payload type 99 could be
1220
m=video 49232 RTP/AVP 99
1221
a=rtpmap:99 X-GSMLPC/8000
1223
Such an experimental encoding requires that any site wishing to
1224
receive the media stream has relevant configured state in its
1225
session directory to know which tools are appropriate.
1227
Note that RTP audio formats typically do not include information
1228
about the number of samples per packet. If a non-default (as
1229
defined in the RTP Audio/Video Profile) packetisation is required,
1230
the "ptime" attribute is used as given below.
1234
Handley & Jacobson Standards Track [Page 22]
1236
RFC 2327 SDP April 1998
1239
For more details on RTP audio and video formats, see [3].
1241
o Formats for non-RTP media should be registered as MIME content
1242
types as described in Appendix B. For example, the LBL whiteboard
1243
application might be registered as MIME content-type application/wb
1244
with encoding considerations specifying that it operates over UDP,
1245
with no appropriate file format. In SDP this would then be
1246
expressed using a combination of the "media" field and the "fmt"
1249
m=application 32416 udp wb
1251
Suggested Attributes
1253
The following attributes are suggested. Since application writers
1254
may add new attributes as they are required, this list is not
1258
This attribute gives the dot-separated hierarchical category of
1259
the session. This is to enable a receiver to filter unwanted
1260
sessions by category. It would probably have been a compulsory
1261
separate field, except for its experimental nature at this time.
1262
It is a session-level attribute, and is not dependent on charset.
1265
Like the cat attribute, this is to assist identifying wanted
1266
sessions at the receiver. This allows a receiver to select
1267
interesting session based on keywords describing the purpose of
1268
the session. It is a session-level attribute. It is a charset
1269
dependent attribute, meaning that its value should be interpreted
1270
in the charset specified for the session description if one is
1271
specified, or by default in ISO 10646/UTF-8.
1273
a=tool:<name and version of tool>
1274
This gives the name and version number of the tool used to create
1275
the session description. It is a session-level attribute, and is
1276
not dependent on charset.
1278
a=ptime:<packet time>
1279
This gives the length of time in milliseconds represented by the
1280
media in a packet. This is probably only meaningful for audio
1281
data. It should not be necessary to know ptime to decode RTP or
1282
vat audio, and it is intended as a recommendation for the
1283
encoding/packetisation of audio. It is a media attribute, and is
1284
not dependent on charset.
1290
Handley & Jacobson Standards Track [Page 23]
1292
RFC 2327 SDP April 1998
1296
This specifies that the tools should be started in receive-only
1297
mode where applicable. It can be either a session or media
1298
attribute, and is not dependent on charset.
1301
This specifies that the tools should be started in send and
1302
receive mode. This is necessary for interactive conferences with
1303
tools such as wb which defaults to receive only mode. It can be
1304
either a session or media attribute, and is not dependent on
1308
This specifies that the tools should be started in send-only
1309
mode. An example may be where a different unicast address is to
1310
be used for a traffic destination than for a traffic source. In
1311
such a case, two media descriptions may be use, one sendonly and
1312
one recvonly. It can be either a session or media attribute, but
1313
would normally only be used as a media attribute, and is not
1314
dependent on charset.
1316
a=orient:<whiteboard orientation>
1317
Normally this is only used in a whiteboard media specification.
1318
It specifies the orientation of a the whiteboard on the screen.
1319
It is a media attribute. Permitted values are `portrait',
1320
`landscape' and `seascape' (upside down landscape). It is not
1321
dependent on charset
1323
a=type:<conference type>
1324
This specifies the type of the conference. Suggested values are
1325
`broadcast', `meeting', `moderated', `test' and `H332'.
1326
`recvonly' should be the default for `type:broadcast' sessions,
1327
`type:meeting' should imply `sendrecv' and `type:moderated'
1328
should indicate the use of a floor control tool and that the
1329
media tools are started so as to "mute" new sites joining the
1332
Specifying the attribute type:H332 indicates that this loosely
1333
coupled session is part of a H.332 session as defined in the ITU
1334
H.332 specification [10]. Media tools should be started
1337
Specifying the attribute type:test is suggested as a hint that,
1338
unless explicitly requested otherwise, receivers can safely avoid
1339
displaying this session description to users.
1341
The type attribute is a session-level attribute, and is not
1342
dependent on charset.
1346
Handley & Jacobson Standards Track [Page 24]
1348
RFC 2327 SDP April 1998
1351
a=charset:<character set>
1352
This specifies the character set to be used to display the
1353
session name and information data. By default, the ISO-10646
1354
character set in UTF-8 encoding is used. If a more compact
1355
representation is required, other character sets may be used such
1356
as ISO-8859-1 for Northern European languages. In particular,
1357
the ISO 8859-1 is specified with the following SDP attribute:
1359
a=charset:ISO-8859-1
1361
This is a session-level attribute; if this attribute is present,
1362
it must be before the first media field. The charset specified
1363
MUST be one of those registered with IANA, such as ISO-8859-1.
1364
The character set identifier is a US-ASCII string and MUST be
1365
compared against the IANA identifiers using a case-insensitive
1366
comparison. If the identifier is not recognised or not
1367
supported, all strings that are affected by it SHOULD be regarded
1370
Note that a character set specified MUST still prohibit the use
1371
of bytes 0x00 (Nul), 0x0A (LF) and 0x0d (CR). Character sets
1372
requiring the use of these characters MUST define a quoting
1373
mechanism that prevents these bytes appearing within text fields.
1375
a=sdplang:<language tag>
1376
This can be a session level attribute or a media level attribute.
1377
As a session level attribute, it specifies the language for the
1378
session description. As a media level attribute, it specifies
1379
the language for any media-level SDP information field associated
1380
with that media. Multiple sdplang attributes can be provided
1381
either at session or media level if multiple languages in the
1382
session description or media use multiple languages, in which
1383
case the order of the attributes indicates the order of
1384
importance of the various languages in the session or media from
1385
most important to least important.
1387
In general, sending session descriptions consisting of multiple
1388
languages should be discouraged. Instead, multiple descriptions
1389
should be sent describing the session, one in each language.
1390
However this is not possible with all transport mechanisms, and
1391
so multiple sdplang attributes are allowed although not
1394
The sdplang attribute value must be a single RFC 1766 language
1395
tag in US-ASCII. It is not dependent on the charset attribute.
1396
An sdplang attribute SHOULD be specified when a session is of
1402
Handley & Jacobson Standards Track [Page 25]
1404
RFC 2327 SDP April 1998
1407
sufficient scope to cross geographic boundaries where the
1408
language of recipients cannot be assumed, or where the session is
1409
in a different language from the locally assumed norm.
1411
a=lang:<language tag>
1412
This can be a session level attribute or a media level attribute.
1413
As a session level attribute, it specifies the default language
1414
for the session being described. As a media level attribute, it
1415
specifies the language for that media, overriding any session-
1416
level language specified. Multiple lang attributes can be
1417
provided either at session or media level if multiple languages
1418
if the session description or media use multiple languages, in
1419
which case the order of the attributes indicates the order of
1420
importance of the various languages in the session or media from
1421
most important to least important.
1423
The lang attribute value must be a single RFC 1766 language tag
1424
in US-ASCII. It is not dependent on the charset attribute. A
1425
lang attribute SHOULD be specified when a session is of
1426
sufficient scope to cross geographic boundaries where the
1427
language of recipients cannot be assumed, or where the session is
1428
in a different language from the locally assumed norm.
1430
a=framerate:<frame rate>
1431
This gives the maximum video frame rate in frames/sec. It is
1432
intended as a recommendation for the encoding of video data.
1433
Decimal representations of fractional values using the notation
1434
"<integer>.<fraction>" are allowed. It is a media attribute, is
1435
only defined for video media, and is not dependent on charset.
1438
This gives a suggestion for the quality of the encoding as an
1441
The intention of the quality attribute for video is to specify a
1442
non-default trade-off between frame-rate and still-image quality.
1443
For video, the value in the range 0 to 10, with the following
1446
10 - the best still-image quality the compression scheme can
1449
5 - the default behaviour given no quality suggestion.
1451
0 - the worst still-image quality the codec designer thinks is
1454
It is a media attribute, and is not dependent on charset.
1458
Handley & Jacobson Standards Track [Page 26]
1460
RFC 2327 SDP April 1998
1463
a=fmtp:<format> <format specific parameters>
1464
This attribute allows parameters that are specific to a
1465
particular format to be conveyed in a way that SDP doesn't have
1466
to understand them. The format must be one of the formats
1467
specified for the media. Format-specific parameters may be any
1468
set of parameters required to be conveyed by SDP and given
1469
unchanged to the media tool that will use this format.
1471
It is a media attribute, and is not dependent on charset.
1473
6.1. Communicating Conference Control Policy
1475
There is some debate over the way conference control policy should be
1476
communicated. In general, the authors believe that an implicit
1477
declarative style of specifying conference control is desirable where
1480
A simple declarative style uses a single conference attribute field
1481
before the first media field, possibly supplemented by properties
1482
such as `recvonly' for some of the media tools. This conference
1483
attribute conveys the conference control policy. An example might be:
1487
In some cases, however, it is possible that this may be insufficient
1488
to communicate the details of an unusual conference control policy.
1489
If this is the case, then a conference attribute specifying external
1490
control might be set, and then one or more "media" fields might be
1491
used to specify the conference control tools and configuration data
1492
for those tools. An example is an ITU H.332 session:
1496
m=audio 49230 RTP/AVP 0
1497
m=video 49232 RTP/AVP 31
1498
m=application 12349 udp wb
1499
m=control 49234 H323 mc
1500
c=IN IP4 134.134.157.81
1502
In this example, a general conference attribute (type:H332) is
1503
specified stating that conference control will be provided by an
1504
external H.332 tool, and a contact addresses for the H.323 session
1505
multipoint controller is given.
1507
In this document, only the declarative style of conference control
1508
declaration is specified. Other forms of conference control should
1509
specify an appropriate type attribute, and should define the
1510
implications this has for control media.
1514
Handley & Jacobson Standards Track [Page 27]
1516
RFC 2327 SDP April 1998
1519
7. Security Considerations
1521
SDP is a session description format that describes multimedia
1522
sessions. A session description should not be trusted unless it has
1523
been obtained by an authenticated transport protocol from a trusted
1524
source. Many different transport protocols may be used to distribute
1525
session description, and the nature of the authentication will differ
1526
from transport to transport.
1528
One transport that will frequently be used to distribute session
1529
descriptions is the Session Announcement Protocol (SAP). SAP
1530
provides both encryption and authentication mechanisms but due to the
1531
nature of session announcements it is likely that there are many
1532
occasions where the originator of a session announcement cannot be
1533
authenticated because they are previously unknown to the receiver of
1534
the announcement and because no common public key infrastructure is
1537
On receiving a session description over an unauthenticated transport
1538
mechanism or from an untrusted party, software parsing the session
1539
should take a few precautions. Session description contain
1540
information required to start software on the receivers system.
1541
Software that parses a session description MUST not be able to start
1542
other software except that which is specifically configured as
1543
appropriate software to participate in multimedia sessions. It is
1544
normally considered INAPPROPRIATE for software parsing a session
1545
description to start, on a user's system, software that is
1546
appropriate to participate in multimedia sessions, without the user
1547
first being informed that such software will be started and giving
1548
their consent. Thus a session description arriving by session
1549
announcement, email, session invitation, or WWW page SHOULD not
1550
deliver the user into an {it interactive} multimedia session without
1551
the user being aware that this will happen. As it is not always
1552
simple to tell whether a session is interactive or not, applications
1553
that are unsure should assume sessions are interactive.
1555
In this specification, there are no attributes which would allow the
1556
recipient of a session description to be informed to start multimedia
1557
tools in a mode where they default to transmitting. Under some
1558
circumstances it might be appropriate to define such attributes. If
1559
this is done an application parsing a session description containing
1560
such attributes SHOULD either ignore them, or inform the user that
1561
joining this session will result in the automatic transmission of
1562
multimedia data. The default behaviour for an unknown attribute is
1570
Handley & Jacobson Standards Track [Page 28]
1572
RFC 2327 SDP April 1998
1575
Session descriptions may be parsed at intermediate systems such as
1576
firewalls for the purposes of opening a hole in the firewall to allow
1577
the participation in multimedia sessions. It is considered
1578
INAPPROPRIATE for a firewall to open such holes for unicast data
1579
streams unless the session description comes in a request from inside
1582
For multicast sessions, it is likely that local administrators will
1583
apply their own policies, but the exclusive use of "local" or "site-
1584
local" administrative scope within the firewall and the refusal of
1585
the firewall to open a hole for such scopes will provide separation
1586
of global multicast sessions from local ones.
1626
Handley & Jacobson Standards Track [Page 29]
1628
RFC 2327 SDP April 1998
1631
Appendix A: SDP Grammar
1633
This appendix provides an Augmented BNF grammar for SDP. ABNF is
1634
defined in RFC 2234.
1637
announcement = proto-version
1651
proto-version = "v=" 1*DIGIT CRLF
1652
;this memo describes version 0
1654
origin-field = "o=" username space
1655
sess-id space sess-version space
1656
nettype space addrtype space
1659
session-name-field = "s=" text CRLF
1661
information-field = ["i=" text CRLF]
1663
uri-field = ["u=" uri CRLF]
1665
email-fields = *("e=" email-address CRLF)
1667
phone-fields = *("p=" phone-number CRLF)
1670
connection-field = ["c=" nettype space addrtype space
1671
connection-address CRLF]
1672
;a connection field must be present
1673
;in every media description or at the
1677
bandwidth-fields = *("b=" bwtype ":" bandwidth CRLF)
1682
Handley & Jacobson Standards Track [Page 30]
1684
RFC 2327 SDP April 1998
1687
time-fields = 1*( "t=" start-time space stop-time
1688
*(CRLF repeat-fields) CRLF)
1689
[zone-adjustments CRLF]
1692
repeat-fields = "r=" repeat-interval space typed-time
1693
1*(space typed-time)
1696
zone-adjustments = time space ["-"] typed-time
1697
*(space time space ["-"] typed-time)
1700
key-field = ["k=" key-type CRLF]
1703
key-type = "prompt" |
1705
"base64:" key-data |
1709
key-data = email-safe | "~" | "
1712
attribute-fields = *("a=" attribute CRLF)
1715
media-descriptions = *( media-field
1723
media-field = "m=" media space port ["/" integer]
1724
space proto 1*(space fmt) CRLF
1727
media = 1*(alpha-numeric)
1728
;typically "audio", "video", "application"
1731
fmt = 1*(alpha-numeric)
1732
;typically an RTP payload type for audio
1738
Handley & Jacobson Standards Track [Page 31]
1740
RFC 2327 SDP April 1998
1743
proto = 1*(alpha-numeric)
1744
;typically "RTP/AVP" or "udp" for IP4
1748
;should in the range "1024" to "65535" inclusive
1749
;for UDP based media
1752
attribute = (att-field ":" att-value) | att-field
1755
att-field = 1*(alpha-numeric)
1758
att-value = byte-string
1762
;should be unique for this originating username/host
1765
sess-version = 1*(DIGIT)
1769
connection-address = multicast-address
1773
multicast-address = 3*(decimal-uchar ".") decimal-uchar "/" ttl
1775
;multicast addresses may be in the range
1776
;224.0.0.0 to 239.255.255.255
1780
start-time = time | "0"
1782
stop-time = time | "0"
1784
time = POS-DIGIT 9*(DIGIT)
1785
;sufficient for 2 more centuries
1788
repeat-interval = typed-time
1794
Handley & Jacobson Standards Track [Page 32]
1796
RFC 2327 SDP April 1998
1799
typed-time = 1*(DIGIT) [fixed-len-time-unit]
1802
fixed-len-time-unit = "d" | "h" | "m" | "s"
1805
bwtype = 1*(alpha-numeric)
1807
bandwidth = 1*(DIGIT)
1811
;pretty wide definition, but doesn't include space
1814
email-address = email | email "(" email-safe ")" |
1815
email-safe "<" email ">"
1818
email = ;defined in RFC822
1821
uri= ;defined in RFC1630
1824
phone-number = phone | phone "(" email-safe ")" |
1825
email-safe "<" phone ">"
1828
phone = "+" POS-DIGIT 1*(space | "-" | DIGIT)
1829
;there must be a space or hyphen between the
1830
;international code and the rest of the number.
1834
;list to be extended
1837
addrtype = "IP4" | "IP6"
1838
;list to be extended
1841
addr = FQDN | unicast-address
1844
FQDN = 4*(alpha-numeric|"-"|".")
1845
;fully qualified domain name as specified in RFC1035
1850
Handley & Jacobson Standards Track [Page 33]
1852
RFC 2327 SDP April 1998
1855
unicast-address = IP4-address | IP6-address
1858
IP4-address = b1 "." decimal-uchar "." decimal-uchar "." b4
1860
;less than "224"; not "0" or "127"
1864
IP6-address = ;to be defined
1868
;default is to interpret this as IS0-10646 UTF8
1869
;ISO 8859-1 requires a "a=charset:ISO-8859-1"
1870
;session-level attribute to be used
1873
byte-string = 1*(0x01..0x09|0x0b|0x0c|0x0e..0xff)
1874
;any byte except NUL, CR or LF
1877
decimal-uchar = DIGIT
1880
| ("2" ("0"|"1"|"2"|"3"|"4") DIGIT)
1881
| ("2" "5" ("0"|"1"|"2"|"3"|"4"|"5"))
1884
integer = POS-DIGIT *(DIGIT)
1887
alpha-numeric = ALPHA | DIGIT
1890
DIGIT = "0" | POS-DIGIT
1893
POS-DIGIT = "1"|"2"|"3"|"4"|"5"|"6"|"7"|"8"|"9"
1896
ALPHA = "a"|"b"|"c"|"d"|"e"|"f"|"g"|"h"|"i"|"j"|"k"|
1897
"l"|"m"|"n"|"o "|"p"|"q"|"r"|"s"|"t"|"u"|"v"|
1898
"w"|"x"|"y"|"z"|"A"|"B"|"C "|"D"|"E"|"F"|"G"|
1899
"H"|"I"|"J"|"K"|"L"|"M"|"N"|"O"|"P"|" Q"|"R"|
1900
"S"|"T"|"U"|"V"|"W"|"X"|"Y"|"Z"
1906
Handley & Jacobson Standards Track [Page 34]
1908
RFC 2327 SDP April 1998
1911
email-safe = safe | space | tab
1914
safe = alpha-numeric |
1915
"'" | "'" | "-" | "." | "/" | ":" | "?" | """ |
1916
"#" | "$" | "&" | "*" | ";" | "=" | "@" | "[" |
1917
"]" | "^" | "_" | "`" | "{" | "|" | "}" | "+" |
1962
Handley & Jacobson Standards Track [Page 35]
1964
RFC 2327 SDP April 1998
1967
Appendix B: Guidelines for registering SDP names with IANA
1969
There are seven field names that may be registered with IANA. Using
1970
the terminology in the SDP specification BNF, they are "media",
1971
"proto", "fmt", "att-field", "bwtype", "nettype" and "addrtype".
1973
"media" (eg, audio, video, application, data).
1975
Packetized media types, such as those used by RTP, share the
1976
namespace used by media types registry [RFC 2048] (i.e. "MIME
1977
types"). The list of valid media names is the set of top-level
1978
MIME content types. The set of media is intended to be small and
1979
not to be extended except under rare circumstances. (The MIME
1980
subtype corresponds to the "fmt" parameter below).
1984
In general this should be an IETF standards-track transport
1985
protocol identifier such as RTP/AVP (rfc 1889 under the rfc 1890
1988
However, people will want to invent their own proprietary
1989
transport protocols. Some of these should be registered as a
1990
"fmt" using "udp" as the protocol and some of which probably
1993
Where the protocol and the application are intimately linked,
1994
such as with the LBL whiteboard wb which used a proprietary and
1995
special purpose protocol over UDP, the protocol name should be
1996
"udp" and the format name that should be registered is "wb". The
1997
rules for formats (see below) apply to such registrations.
1999
Where the proprietary transport protocol really carries many
2000
different data formats, it is possible to register a new protocol
2001
name with IANA. In such a case, an RFC MUST be produced
2002
describing the protocol and referenced in the registration. Such
2003
an RFC MAY be informational, although it is preferable if it is
2008
The format namespace is dependent on the context of the "proto"
2009
field, so a format cannot be registered without specifying one or
2010
more transport protocols that it applies to.
2012
Formats cover all the possible encodings that might want to be
2013
transported in a multimedia session.
2018
Handley & Jacobson Standards Track [Page 36]
2020
RFC 2327 SDP April 1998
2023
For RTP formats that have been assigned static payload types, the
2024
payload type number is used. For RTP formats using a dynamic
2025
payload type number, the dynamic payload type number is given as
2026
the format and an additional "rtpmap" attribute specifies the
2027
format and parameters.
2029
For non-RTP formats, any unregistered format name may be
2030
registered through the MIME-type registration process [RFC 2048].
2031
The type given here is the MIME subtype only (the top-level MIME
2032
content type is specified by the media parameter). The MIME type
2033
registration SHOULD reference a standards-track RFC which
2034
describes the transport protocol for this media type. If there
2035
is an existing MIME type for this format, the MIME registration
2036
should be augmented to reference the transport specification for
2037
this media type. If there is not an existing MIME type for this
2038
format, and there exists no appropriate file format, this should
2039
be noted in the encoding considerations as "no appropriate file
2042
"att-field" (Attribute names)
2044
Attribute field names MAY be registered with IANA, although this
2045
is not compulsory, and unknown attributes are simply ignored.
2047
When an attribute is registered, it must be accompanied by a
2048
brief specification stating the following:
2050
o contact name, email address and telephone number
2052
o attribute-name (as it will appear in SDP)
2054
o long-form attribute name in English
2056
o type of attribute (session level, media level, or both)
2058
o whether the attribute value is subject to the charset
2061
o a one paragraph explanation of the purpose of the attribute.
2063
o a specification of appropriate attribute values for this
2066
IANA will not sanity check such attribute registrations except to
2067
ensure that they do not clash with existing registrations.
2074
Handley & Jacobson Standards Track [Page 37]
2076
RFC 2327 SDP April 1998
2079
Although the above is the minimum that IANA will accept, if the
2080
attribute is expected to see widespread use and interoperability
2081
is an issue, authors are encouraged to produce a standards-track
2082
RFC that specifies the attribute more precisely.
2084
Submitters of registrations should ensure that the specification
2085
is in the spirit of SDP attributes, most notably that the
2086
attribute is platform independent in the sense that it makes no
2087
implicit assumptions about operating systems and does not name
2088
specific pieces of software in a manner that might inhibit
2091
"bwtype" (bandwidth specifiers)
2093
A proliferation of bandwidth specifiers is strongly discouraged.
2095
New bandwidth specifiers may be registered with IANA. The
2096
submission MUST reference a standards-track RFC specifying the
2097
semantics of the bandwidth specifier precisely, and indicating
2098
when it should be used, and why the existing registered bandwidth
2099
specifiers do not suffice.
2101
"nettype" (Network Type)
2103
New network types may be registered with IANA if SDP needs to be
2104
used in the context of non-internet environments. Whilst these
2105
are not normally the preserve of IANA, there may be circumstances
2106
when an Internet application needs to interoperate with a non-
2107
internet application, such as when gatewaying an internet
2108
telephony call into the PSTN. The number of network types should
2109
be small and should be rarely extended. A new network type
2110
cannot be registered without registering at least one address
2111
type to be used with that network type. A new network type
2112
registration MUST reference an RFC which gives details of the
2113
network type and address type and specifies how and when they
2114
would be used. Such an RFC MAY be Informational.
2116
"addrtype" (Address Type)
2118
New address types may be registered with IANA. An address type
2119
is only meaningful in the context of a network type, and any
2120
registration of an address type MUST specify a registered network
2121
type, or be submitted along with a network type registration. A
2122
new address type registration MUST reference an RFC giving
2123
details of the syntax of the address type. Such an RFC MAY be
2124
Informational. Address types are not expected to be registered
2130
Handley & Jacobson Standards Track [Page 38]
2132
RFC 2327 SDP April 1998
2135
Registration Procedure
2137
To register a name the above guidelines should be followed regarding
2138
the required level of documentation that is required. The
2139
registration itself should be sent to IANA. Attribute registrations
2140
should include the information given above. Other registrations
2141
should include the following additional information:
2143
o contact name, email address and telephone number
2145
o name being registered (as it will appear in SDP)
2147
o long-form name in English
2149
o type of name ("media", "proto", "fmt", "bwtype", "nettype", or
2152
o a one paragraph explanation of the purpose of the registered name.
2154
o a reference to the specification (eg RFC number) of the registered
2157
IANA may refer any registration to the IESG or to any appropriate
2158
IETF working group for review, and may request revisions to be made
2159
before a registration will be made.
2186
Handley & Jacobson Standards Track [Page 39]
2188
RFC 2327 SDP April 1998
2191
Appendix C: Authors' Addresses
2194
Information Sciences Institute
2195
c/o MIT Laboratory for Computer Science
2196
545 Technology Square
2199
electronic mail: mjh@isi.edu
2203
Lawrence Berkeley Laboratory
2206
electronic mail: van@ee.lbl.gov
2210
Many people in the IETF MMUSIC working group have made comments and
2211
suggestions contributing to this document. In particular, we would
2212
like to thank Eve Schooler, Steve Casner, Bill Fenner, Allison
2213
Mankin, Ross Finlayson, Peter Parnes, Joerg Ott, Carsten Bormann, Rob
2214
Lanphier and Steve Hanna.
2218
[1] Mills, D., "Network Time Protocol (version 3) specification and
2219
implementation", RFC 1305, March 1992.
2221
[2] Schulzrinne, H., Casner, S., Frederick, R. and V. Jacobson, "RTP:
2222
A Transport Protocol for Real-Time Applications", RFC 1889, January
2225
[3] Schulzrinne, H., "RTP Profile for Audio and Video Conferences
2226
with Minimal Control", RFC 1890, January 1996
2228
[4] Handley, M., "SAP - Session Announcement Protocol", Work in
2231
[5] V. Jacobson, S. McCanne, "vat - X11-based audio teleconferencing
2232
tool" vat manual page, Lawrence Berkeley Laboratory, 1994.
2234
[6] The Unicode Consortium, "The Unicode Standard -- Version 2.0",
2235
Addison-Wesley, 1996.
2242
Handley & Jacobson Standards Track [Page 40]
2244
RFC 2327 SDP April 1998
2247
[7] ISO/IEC 10646-1:1993. International Standard -- Information
2248
technol- ogy -- Universal Multiple-Octet Coded Character Set (UCS) --
2249
Part 1: Architecture and Basic Multilingual Plane. Five amendments
2250
and a techn- ical corrigendum have been published up to now. UTF-8
2251
is described in Annex R, published as Amendment 2.
2253
[8] Goldsmith, D., and M. Davis, "Using Unicode with MIME", RFC 1641,
2256
[9] Yergeau, F., "UTF-8, a transformation format of Unicode and ISO
2257
10646", RFC 2044, October 1996.
2259
[10] ITU-T Recommendation H.332 (1998): "Multimedia Terminal for
2260
Receiving Internet-based H.323 Conferences", ITU, Geneva.
2262
[11] Handley, M., Schooler, E., and H. Schulzrinne, "Session
2263
Initiation Protocol (SIP)", Work in Progress.
2265
[12] Schulzrinne, H., Rao, A., and R. Lanphier, "Real Time Streaming
2266
Protocol (RTSP)", RFC 2326, April 1998.
2298
Handley & Jacobson Standards Track [Page 41]
2300
RFC 2327 SDP April 1998
2303
Full Copyright Statement
2305
Copyright (C) The Internet Society (1998). All Rights Reserved.
2307
This document and translations of it may be copied and furnished to
2308
others, and derivative works that comment on or otherwise explain it
2309
or assist in its implementation may be prepared, copied, published
2310
and distributed, in whole or in part, without restriction of any
2311
kind, provided that the above copyright notice and this paragraph are
2312
included on all such copies and derivative works. However, this
2313
document itself may not be modified in any way, such as by removing
2314
the copyright notice or references to the Internet Society or other
2315
Internet organizations, except as needed for the purpose of
2316
developing Internet standards in which case the procedures for
2317
copyrights defined in the Internet Standards process must be
2318
followed, or as required to translate it into languages other than
2321
The limited permissions granted above are perpetual and will not be
2322
revoked by the Internet Society or its successors or assigns.
2324
This document and the information contained herein is provided on an
2325
"AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING
2326
TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING
2327
BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION
2328
HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF
2329
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
2354
Handley & Jacobson Standards Track [Page 42]