One document matched: draft-ietf-ice-trickle-00.xml
<?xml version="1.0" encoding="UTF-8"?>
<?xml-stylesheet type='text/xsl' href='rfc2629.xslt' ?>
<!DOCTYPE rfc SYSTEM "rfc2629.dtd">
<rfc category='std' ipr='trust200902'
docName='draft-ietf-ice-trickle-00'>
<?rfc toc='yes' ?>
<?rfc symrefs='yes' ?>
<?rfc sortrefs='yes'?>
<?rfc iprnotified='no' ?>
<?rfc strict='yes' ?>
<?rfc compact='yes' ?>
<front>
<title abbrev='Trickle ICE'>
Trickle ICE: Incremental Provisioning of Candidates for the
Interactive Connectivity Establishment (ICE) Protocol
</title>
<author initials='E.' surname='Ivov'
fullname='Emil Ivov'>
<organization abbrev='Jitsi'>Jitsi</organization>
<address>
<postal>
<street></street>
<city>Strasbourg</city>
<code>67000</code>
<country>France</country>
</postal>
<phone>+33 6 72 81 15 55</phone>
<email>emcho@jitsi.org</email>
</address>
</author>
<author fullname="Eric Rescorla" initials="E.K." surname="Rescorla">
<organization>RTFM, Inc.</organization>
<address>
<postal>
<street>2064 Edgewood Drive</street>
<city>Palo Alto</city>
<region>CA</region>
<code>94303</code>
<country>USA</country>
</postal>
<phone>+1 650 678 2350</phone>
<email>ekr@rtfm.com</email>
</address>
</author>
<author fullname="Justin Uberti" initials="J." surname="Uberti">
<organization>Google</organization>
<address>
<postal>
<street>747 6th St S</street>
<city>Kirkland</city>
<region>WA</region>
<code>98033</code>
<country>USA</country>
</postal>
<phone>+1 857 288 8888</phone>
<email>justin@uberti.name</email>
</address>
</author>
<author initials="P." surname="Saint-Andre" fullname="Peter Saint-Andre">
<organization>&yet</organization>
<address>
<email>peter@andyet.com</email>
<uri>https://andyet.com/</uri>
</address>
</author>
<date />
<abstract>
<t>
This document describes an extension to the Interactive
Connectivity Establishment (ICE) protocol that allows ICE agents
to send and receive candidates incrementally rather than
exchanging complete lists. With such incremental provisioning,
ICE agents can begin connectivity checks while they are still
gathering candidates and considerably shorten the time necessary
for ICE processing to complete. This mechanism is called "trickle
ICE".
</t>
</abstract>
</front>
<middle>
<section title='Introduction'>
<t>
The Interactive Connectivity Establishment (ICE) protocol
<xref target="RFC5245"/> describes mechanisms for gathering
candidates, prioritizing them, choosing default ones, exchanging
them with the remote party, pairing them and ordering them into
check lists. Once all of the above have been completed, and only
then, the participating agents can begin a phase of connectivity
checks and eventually select the pair of candidates that will be
used in the following session.
</t>
<t>
While the above sequence has the advantage of being relatively
straightforward to implement and debug once deployed, it may
also prove to be rather lengthy. Gathering candidates or
candidate gathering often involves things like querying
<xref target="RFC5389">STUN</xref> servers, discovering UPnP
devices, and allocating relayed candidates at
<xref target="RFC5766">TURN</xref> servers. All of these can
be delayed for a noticeable amount of time and while they can be
run in parallel, they still need to respect the pacing
requirements from <xref target="RFC5245"/>, which is likely to
delay them even further. Some or all of the above would also
have to be completed by the remote agent. Both agents would
next perform connectivity checks and only then would they be
ready to begin streaming media.
</t>
<t>
All of the above can lead to relatively lengthy session
establishment times and degraded user experience.
</t>
<t>
The purpose of this document is to define an alternative mode of
operation for ICE implementations, also known as "trickle ICE",
where candidates can be exchanged incrementally. This would
allow ICE agents to exchange candidates as soon as a
session has been initiated. Connectivity checks for a media
stream would also start as soon as the first candidates for that
stream have become available.
</t>
<t>
Trickle ICE allows reducing session establishment times in cases
where connectivity is confirmed for the first exchanged
candidates (e.g. where the host candidates for one of the agents
are directly reachable from the second agent). Even when this is
not the case, running candidate gathering for both agents and
connectivity checks all in parallel allows to considerably
reduce ICE processing times.
</t>
<t>
It is worth pointing out that before being introduced to the
IETF, trickle ICE had already been included in specifications
such as <xref target="XEP-0176">XMPP Jingle</xref> and it has
been in use in various implementations and deployments.
</t>
<t>
In addition to the basics of trickle ICE, this document also
describes how to discover support for trickle ICE,
how regular ICE processing needs to be modified when
building and updating check lists, and how trickle ICE
implementations interoperate with agents that only
implement <xref target="RFC5245"/> processing.
</t>
<t>
This specification does not define usage of trickle ICE with any
specific signalling protocol, different from
<xref target="RFC5245"/> which contains a usage for ICE with
SIP <xref target='RFC3261'/>. Such usages would have to be specified in separate
documents such as for example
<xref target="I-D.ietf-mmusic-trickle-ice-sip"/>. However,
trickle ICE does however reuse and build upon the SDP syntax
defined by <xref target="RFC5245"/>.
</t>
<t>
Although this document mostly describes trickle ICE in terms
of the offer/answer model <xref target='RFC3264'/>, trickle
ICE (and ICE itself) can be used by application protocols that
do not follow the offer/answer model.
</t>
</section>
<section title="Terminology">
<t>
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL
NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and
"OPTIONAL" in this document are to be interpreted as described
in <xref target="RFC2119"/>.
</t>
<t>
This specification makes use of all terminology defined by the
protocol for Interactive Connectivity Establishment in
<xref target="RFC5245"/>.
</t>
<t>
<list style="hanging">
<t hangText="Vanilla ICE:">
The Interactive Connectivity Establishment protocol as
defined in <xref target="RFC5245"/>. Through the rest of the
text, the terms vanilla ICE and "RFC5245" are used
interchangeably.
</t>
<t hangText="Candidate Harvester:">
A module used by an ICE agent to obtain local candidates.
Candidate gatherers use different mechanisms for
discovering local candidates. Some of them would typically
make use of protocols such as STUN or TURN. Others may also
employ techniques that are not referenced within
<xref target="RFC5245"/>. UPnP based port allocation and
XMPP Jingle Relay Nodes <xref target="XEP-0278"/> are among
the possible examples.
</t>
<t hangText="Trickled Candidates:">
Candidates that a trickle ICE agent is sending subsequently
to but within the context defined by an offer or an answer.
Trickled candidates can be sent in parallel with candidate
gathering and connectivity checks.
</t>
<t hangText="Trickling/Trickle (v.):">
The act of sending trickled candidates.
</t>
<t hangText="Half Trickle:">
A trickle ICE mode of operation where the offerer gathers
its first generation of candidates strictly before creating
and sending the offer. Once sent, that offer can be
processed by vanilla ICE agents and does not require support
for this specification. It also allows trickle ICE capable
answerers to still gather candidates and perform
connectivity checks in a non-blocking way, thus roughly
offering "half" the advantages of trickle ICE. The mechanism
is mostly meant for use in cases where support for trickle
ICE cannot be confirmed prior to sending a initial offer.
</t>
<t hangText="Full Trickle:">
Regular mode of operation for trickle ICE agents, used in
opposition to the half trickle mode of operation.
</t>
</list>
</t>
</section>
<section title='Determining Support for Trickle ICE' anchor="disco">
<t>
According to <xref target="RFC5245"/>, supported features are to
be advertised in the ice-options attribute. Therefore an agent
supporting trickle ICE MUST include a token of "trickle" in the
ice-options attribute every time it generates an offer or an answer.
Syntax for this token is defined in <xref target="sdp.offer"/>.
</t>
<t>
Agents that receive offers or answers can verify support by
examining them for the "trickle" ice-options token. However,
agents that are about to send an initial offer have no way of
doing this. Thus usages of trickle for specific protocols need
to either:
</t>
<t>
<list style='symbols'>
<t>
Provide a way for agents to verify support of trickle ICE
prior to initiating a session (XMPP's
<xref target="XEP-0030">Service Discovery</xref> is an
example of one such mechanism); or
</t>
<t>
Make support for trickle ICE mandatory so that support could
be assumed the agents.
</t>
</list>
</t>
<t>
Alternately, for cases where a protocol provides neither of the
above, agents may either rely on provisioning/configuration, or
use the half trickle procedure described in
<xref target="half-trickle"/>.
</t>
<t>
Prior to sending an initial offer, agents using signaling protocols
that support capabilities discovery MAY attempt to verify whether or
not the remote party supports trickle ICE. If an agent determines
that the remote party does not support trickle ICE, it MUST fall back
to using vanilla ICE or abandon the entire session.
</t>
<t>
All trickle ICE offers and answers MUST indicate support of this
specification, as explained in <xref target="sdp.offer"/>.
</t>
<t>
Note that out-of-band discovery semantics and half trickle are
only necessary prior to session initiation, or in other words,
when sending the initial offer. Once a session is established
and trickle ICE support is confirmed for both parties, either
agent can use full trickle for subsequent offers.
</t>
</section>
<section title='Sending the Initial Offer' anchor="initial-offer">
<t>
An agent starts gathering candidates as soon as it has an
indication that communication is imminent (e.g. a user interface
cue or an explicit request to initiate a session). Contrary to
vanilla ICE, implementations of trickle ICE do not need to
gather candidates in a blocking manner. Therefore, unless half
trickle is being used, agents SHOULD generate and transmit their
initial offer as early as possible, in order to allow the remote
party to start gathering and trickling candidates.
</t>
<t>
Trickle ICE agents MAY include any set of candidates in an
offer. This includes the possibility of generating one with no
candidates, or one that contains all the candidates that the
agent is planning on using in the following session.
</t>
<t>
For optimal performance, it is RECOMMENDED that the candidates in
an initial offer (if any) be host candidates only. This would allow
both agents to start gathering server reflexive, relayed and other
non-host candidates simultaneously, and it would also enable them to
begin connectivity checks.
</t>
<t>
If the privacy implications of revealing host addresses are a
concern, agents MAY generate an offer that contains no
candidates and then only trickle candidates that do not reveal
host addresses (e.g. relayed candidates).
</t>
<t>
Methods for calculating priorities and foundations, as well as
determining redundancy of candidates, work just as with vanilla
ICE.
</t>
<section title='Encoding the SDP' anchor="sdp.offer">
<t>
The process of encoding the SDP <xref target="RFC4566"/> is
mostly the same as the one used by vanilla ICE. Still, trickle
ICE does require a few differences described here.
</t>
<t>
Agents MUST indicate support for Trickle ICE by including the
"trickle" token for the "a=ice-options" attribute:
<figure>
<artwork>
<![CDATA[
a=ice-options:trickle
]]>
</artwork>
</figure>
</t>
<t>
As mentioned earlier in this section, offers and answers can
contain any set of candidates, which means that a trickle ICE
session description MAY contain no candidates at all. Doing so
enables the offerer to receive the answerer's initial candidate
list sooner, and also enables the answerer to begin candidate
gathering more quickly. In such
cases the agent would still need to place an address in the
"c=" line(s). If the use of a host address there is
undesirable (e.g., for privacy reasons), the agent MAY set the
connection address to 0.0.0.0. In this case it MUST also
set the port number to 9 (Discard). There is no need to
include a fictitious candidate for the 0.0.0.0 address when
doing so.
</t>
<t>
It is worth noting that the use of IP6 :: has been selected
over IP4 0.0.0.0, even though <xref target="RFC3264"/> already
gives the latter semantics appropriate for such use. The
reason for this choice is the historic use of 0.0.0.0 as a
means of putting a stream on hold <xref target="RFC2543"/> and
the ambiguity that this may cause with legacy libraries and
applications.
</t>
<t>
It is also worth mentioning that use of IP6 :: here does not
constitute any kind of indication as to the actual use of
IPv6 candidates in a session and it can very well appear in
a negotiation that only involves IPv4 candidates.
</t>
</section>
</section>
<section title='Receiving the Initial Offer' >
<t>
When an agent receives an initial offer, it will first check if
it indicates support for trickle ICE as explained in
<xref target="disco"/>. If this is not the case, the agent MUST
process the offer according to the <xref target="RFC5245"/>
procedures or standard <xref target="RFC3264"/> processing in
case no ICE support is detected at all.
</t>
<t>
It is worth pointing out that in case support for trickle ICE is
confirmed, an agent will automatically assume support for
vanilla ICE as well even if the support verification procedure
in <xref target="RFC5245"/> indicates otherwise. Specifically,
the rules from RFC 5245 would imply that ICE itself is not
supported if the initial offer includes no candidates in the
offer; however, such a conclusion is not warranted if the answerer
can confirm that the offerer supports trickle ICE. In this case,
the IP6 :: address present in the "c=" line would not "appear in
a candidate attribute". Fallback to <xref target="RFC3264"/> is not
necessary in this scenario.
</t>
<t>
If, the offer does indicate support for trickle ICE, the agent
will determine its role, start gathering and prioritizing
candidates and, while doing so it will also respond by sending
its own answer, so that both agents can start forming check
lists and begin connectivity checks.
</t>
<section title="Sending the Initial Answer">
<t>
An agent can respond to an initial offer at any point while
gathering candidates. The answer can again contain any set of
candidates, including all candidates or no candidates. (The
benefit of including no candidates is to send the answer as
quickly as possible, so that both parties can consider the
overall session to be under active negotiation as soon as
possible.) Unless the answering agent is
protecting host addresses for privacy reasons, it would
typically construct this initial answer including only them,
thus allowing the remote party to also start forming
checklists and performing connectivity checks.
</t>
<t>
The answer MUST indicate support for trickle ICE as described
by <xref target="disco"/>.
</t>
</section>
<section title="Forming Check Lists and Beginning Connectivity
Checks" anchor="check.lists">
<t>
After exchanging offer and answer, and as soon as they have
obtained local and remote candidates, agents will begin
forming candidate pairs, computing their priorities and
creating check lists according to the vanilla ICE procedures
described in <xref target="RFC5245"/>. Obviously in order for
candidate pairing to be possible, candidates would need to be
provided in both the offer and the answer. If not, then the
agents will still create the check lists (so
that their Active/Frozen state could be monitored and updated)
but they will only populate the check lists once they actually
have the candidate pairs.
</t>
<t>
Initially, all check lists will have their Active/Frozen state
set to Frozen.
</t>
<t>
Trickle ICE agents will then inspect the first check list and
attempt to unfreeze all candidates belonging to the first
component on the first media stream (i.e. the first media
stream that was reported to the ICE implementation from the
using application). If this checklist is still empty however,
agents will hold off further processing until this is no
longer the case.
</t>
<t>
Respecting the order in which lists have been reported to an
ICE implementation, or in other words, the order in which
they appear in SDP, is crucial to the frozen candidates
algorithm and important when making sure that connectivity
checks are performed simultaneously by both agents.
</t>
</section>
<section title='Encoding the SDP' anchor="sdp.answer">
<t>
The process for encoding the SDP at the answerer is identical
to the process followed by the offerer for both full and lite
implementations, as described in <xref target="sdp.offer"/>.
</t>
</section>
</section>
<section title="Receiving the Initial Answer">
<t>
When receiving an answer, agents will follow vanilla ICE
procedures to determine their role and they would then
form check lists (as described in <xref target="check.lists"/>)
and begin connectivity checks .
</t>
</section>
<section title='Performing Connectivity Checks' >
<t>
For the most part, trickle ICE agents perform connectivity
checks following vanilla ICE procedures. Of course, the
asynchronous nature of gathering and communicating candidates
in trickle ICE would impose a number of changes described here.
</t>
<section title="Scheduling Checks" anchor="scheduling-checks">
<t>
The ICE specification <xref target='RFC5245'/>, Section 5.8,
requires that agents terminate the timer for a triggered
check in relation to an active check list once the agent
has exhausted all frozen pairs in check list. This will
not work with trickle ICE, because more pairs will be
added to the check list incrementally.
</t>
<t>
Therefore, a trickle ICE agent SHOULD NOT terminate the timer
until the state of the check list is completed or failed as
specified herein (see <xref target='end-of-candidates'/>).
</t>
</section>
<section title="Check List and Timer State Updates"
anchor="state-updates">
<t>
The ICE specification <xref target='RFC5245'/>, Section 7.1.3.3,
requires that agents update check lists and timer states upon
completing a connectivity check transaction. During such an
update vanilla ICE agents would set the state of a check list
to Failed if both of the following two conditions are satisfied:
</t>
<t>
<list style="symbols">
<t>
all of the pairs in the check list are either in the
Failed or Succeeded state; and
</t>
<t>
there is not a pair in the valid list for each component
of the media stream.
</t>
</list>
</t>
<t>
With trickle ICE, the above situation would often occur when
candidate gathering and trickling are still in progress, even
though it is quite possible that future checks will succeed. For
this reason trickle ICE agents add the following conditions to
the above list:
</t>
<t>
<list style="symbols">
<t>
all candidate gatherers have completed and the agent
is not expecting to discover any new local candidates;
</t>
<t>
the remote agent has sent an end-of-candidates indication
for that check list as described in
<xref target="end-of-candidates"/>.
</t>
</list>
</t>
<t>
Vanilla ICE requires that agents then update all other check
lists, placing one pair in each of them into the Waiting
state, effectively unfreezing all remaining check lists. Given
that with trickle ICE, other check lists may still be empty at
that point, a trickle ICE agent SHOULD also maintain an
explicit Active/Frozen state for every check list, rather than
deducing it from the state of the pairs it contains. This
state should be set to Active when unfreezing the first pair
in a list or when that couldn't happen because a list was
empty.
</t>
</section>
</section>
<section title='Discovering and Sending Additional Local Candidates'
anchor="send-trickling">
<t>
After an offer or an answer have been sent, agents will most
likely continue discovering new local candidates as STUN, TURN
and other non-host candidate gathering mechanisms begin to
yield results. Whenever an agent discovers such a new candidate
it will compute its priority, type, foundation and component id
according to normal vanilla ICE procedures.
</t>
<t>
The new candidate is then checked for redundancy against the
existing list of local candidates. If its transport address and
base match those of an existing candidate, it will be considered
redundant and will be ignored. This would often happen for
server reflexive candidates that match the host addresses they
were obtained from (e.g. when the latter are public IPv4
addresses). Contrary to vanilla ICE, trickle ICE agents will
consider the new candidate redundant regardless of its priority.
</t>
<t>
Next the client sends (i.e. trickles) the newly learnt
candidate(s) to the remote agent. The actual delivery of the new
candidates will be specified by using protocols such as SIP.
Trickle ICE imposes no restrictions on the way this is done or
whether it is done at all. For example, some applications may
choose not to send trickle updates for server reflexive
candidates and rely on the discovery of peer reflexive ones
instead.
</t>
<t>
When trickle updates are sent however, each candidate MUST be
delivered to the receiving Trickle ICE implementation not more
than once and in the same order that they were sent. In other
words, if there are any candidate retransmissions, they must
be hidden from the ICE implementation.
</t>
<t>
Also, candidate trickling needs to be correlated to a specific
ICE negotiation session, so that if there is an ICE restart, any
delayed updates for a previous session can be recognized as such
and ignored by the receiving party.
</t>
<t>
One important aspect of Vanilla ICE is that connectivity checks
for a specific foundation and component be attempted
simultaneously by both agents, so that any firewalls or NATs
fronting the agents would whitelist both endpoints and allow
all except for the first (suicide) packets to go through. This
is also crucial to unfreezing candidates in the right time.
</t>
<t>
In order to preserve this feature here, when trickling
candidates agents MUST respect the order of the components as
they appear (implicitly or explicitly) in the Offer/Answer
descriptions. Therefore a candidate for a specific component
MUST NOT be sent prior to candidates for other components within
the same foundation.
</t>
<t>
For example, the following session description contains two
components (RTP and RTCP), and two foundations (host and the
server reflexive):
<figure>
<artwork>
<![CDATA[
v=0
o=jdoe 2890844526 2890842807 IN IP4 10.0.1.1
s=
c=IN IP4 10.0.1.1
t=0 0
a=ice-pwd:asd88fgpdd777uzjYhagZg
a=ice-ufrag:8hhY
m=audio 5000 RTP/AVP 0
a=rtpmap:0 PCMU/8000
a=candidate:1 1 UDP 2130706431 10.0.1.1 5000 typ host
a=candidate:1 2 UDP 2130706431 10.0.1.1 5001 typ host
a=candidate:2 1 UDP 1694498815 192.0.2.3 5000 typ srflx
raddr 10.0.1.1 rport 8998
a=candidate:2 2 UDP 1694498815 192.0.2.3 5001 typ srflx
raddr 10.0.1.1 rport 8998
]]>
</artwork>
</figure>
For this description the RTCP host candidate MUST NOT be sent
prior to the RTP host candidate. Similarly the RTP server
reflexive candidate MUST be sent together with or prior to the
RTCP server reflexive candidate.
</t>
<t>
Note that the order restriction only applies among candidates
that belong to the same foundation.
</t>
<t>
It is also equally important to preserve this order across media
streams and this is covered by the requirement to always start
unfreezing candidates starting from the first media stream
<xref target="check.lists"/>.
</t>
<t>
Once the candidate has been sent to the remote party, the agent
checks if any remote candidates are currently known for this
same stream. If this is not the case the new candidate will
simply be added to the list of local candidates.
</t>
<t>
Otherwise, if the agent has already learned of one or more
remote candidates for this stream and component, it will begin
pairing the new local candidates with them and adding the pairs
to the existing check lists according to their priority.
</t>
<section title='Pairing Newly Learned Candidates and Updating
Check Lists'
anchor="cand-pairing">
<t>
Forming candidate pairs will work the way it is described by
the ICE specification <xref target='RFC5245'/>. Actually adding the new pair to
a check list however, will happen according to the rules
described below.
</t>
<t>
If the check list where the pair is to be added already
contains the maximum number of candidate pairs (100 by default
as per <xref target="RFC5245"/>), the new pair is discarded.
</t>
<t>
If the new pair's local candidate is server reflexive, the
server reflexive candidate MUST be replaced by its base before
adding the pair to the list. Once this is done, the agent
examines the check list looking for another pair that would be
redundant with the new one. If such a pair exists, the newly
formed pair is ignored.
</t>
<t>
For all other pairs, including those with a server reflexive
local candidate that were not found to be redundant:
<list style="symbols">
<t>
if this check list is Frozen then the new pair will
also be assigned a Frozen state.
</t>
<t>
else if the check list is Active and it is either empty or
contains only candidates in the Succeeded and Failed
states, then the new pair's state is set to Waiting.
</t>
<t>
else if the check list is non-empty and Active, then the
new pair state will be set to
<list style="hanging">
<t hangText="Frozen: ">
if there is at least one pair in the list whose
foundation matches the one in the new pair and whose
state is neither Succeeded nor Failed (eventually the
new pair will get unfrozen after the the on-going
check for the existing pair concludes);
</t>
<t hangText="Waiting: ">
if the list contains no pairs with the same foundation
as the new one, or, in case such pairs exist but they
are all in either the Succeeded or Failed states.
</t>
</list>
</t>
</list>
</t>
</section>
<section title="Encoding the SDP for Additional Candidates">
<t>
To facilitate interoperability an ICE agent will encode
additional candidates using the vanilla ICE SDP syntax. For
example:
<figure>
<artwork>
<![CDATA[
a=candidate:2 1 UDP 1658497328 198.51.100.33 5000 typ host
]]>
</artwork>
</figure>
Given that such lines do not provide a relationship between
the candidate and the m line that it relates to, signalling
protocols using trickle ICE MUST establish that relation
themselves using an <xref target="RFC3388">MID</xref>. Such
MIDs use "media stream identification", as defined in
<xref target="RFC3388"/>, to identify a corresponding m-line.
When creating candidate lines usages of trickle ICE MUST use
the MID if possible, or the m-line index if not. Obviously,
agents MUST NOT send individual candidates prior to generating
the corresponding SDP session description.
</t>
<t>
The exact means of transporting additional candidates to a
remote agent is left to the protocols using trickle ICE. It is
important to note, however, that these candidate exchanges are
not part of the offer/answer model.
</t>
</section>
<section title='Announcing End of Candidates'
anchor="end-of-candidates">
<t>
Once all candidate gatherers for a specific media stream
complete, or expire, the agents will generate an
"end-of-candidates" indication for that stream and send it to
the remote agent via the signalling channel. Such indications
are sent in the form of a media-level attribute that has the
following form: end-of-candidates.
<figure>
<artwork>
<![CDATA[
a=end-of-candidates
]]>
</artwork>
</figure>
The end-of-candidates indications can be sent in the following
ways:
<list style='symbols'>
<t>As part of an offer (which would typically be the case with
half trickle initial offers)</t>
<t>Along with the last candidate an agent can send for a stream</t>
<t>As a standalone notification (e.g., after STUN Binding requests
or TURN Allocate requests to a server timeout and the agent has
no other active gatherers)</t>
</list>
</t>
<t>
Controlled trickle ICE agents SHOULD always send
end-of-candidates indications once gathering for a media stream
has completed unless ICE processing terminates before they've
had a chance to do so. Sending the indication is necessary in
order to avoid ambiguities and speed up ICE conclusion.
Controlling agents on the other hand MAY sometimes
conclude ICE processing prior to sending end-of-candidates
notifications for all streams. This would typically be the case
with aggressive nomination. Yet it is RECOMMENDED that
controlling agents do send such indications whenever possible
for the sake of consistency and keeping middle boxes and
controlled agents up-to-date on the state of ICE processing.
</t>
<t>
When sending end-of-candidates during trickling, rather than as
a part of an offer or an answer, it is the responsibility of the
using protocol to define means that can be used to relate the
indication to one or more specific m-lines.
</t>
<t>
Receiving an end-of-candidates notification allows an agent to
update check list states and, in case valid pairs do not exist
for every component in every media stream, determine that ICE
processing has failed. It also allows agents to speed ICE
conclusion in cases where a candidate pair has been validates
but it involves the use of lower-preference transports such as
TURN. In such situations some implementations may choose to wait
in case higher-priority candidates are received and
end-of-candidates provides an indication that this is not going
to happen.
</t>
<t>
An agent MAY also choose to generate an end-of-candidates
event before candidate gathering has actually completed, if the
agent determines that gathering has continued for more than an
acceptable period of time. However, an agent MUST NOT send any
more candidates after it has send an end-of-candidates
notification.
</t>
<t>
When performing half trickle agents SHOULD send
end-of-candidates together with their initial offer unless they
are planning on potentially sending additional candidates in
case the remote party turns out to actually support trickle ICE.
</t>
<t>
When end-of-candidates is sent as part of an offer or an answer
it can appear as a session-level attribute, which would be
equivalent to having it appear in all m-lines.
</t>
<t>
Once an agent sends the end-of-candidates event, it will
update the state of the corresponding check list as explained
in <xref target="state-updates"/>. Past that point
agents MUST NOT send any new candidates within this ICE session.
Once an agent has received an end-of-candidates indication, it
MUST also ignore any newly received candidates for that media
stream, and adding new candidates to the negotiation is only
possible through an ICE restart.
</t>
<t>
This specification does not
override vanilla ICE semantics for concluding ICE processing.
Therefore even if end-of-candidates indications are sent
agents will still have to go through pair nomination. Also, if
pairs have been nominated for components and media streams, ICE
processing will still conclude even if end-of-candidate
indications have not been received for all streams.
</t>
</section>
</section>
<section title='Receiving Additional Remote Candidates'
anchor="recv-trickling">
<t>
At any point of ICE processing, a trickle ICE agent may receive
new candidates from the remote agent. When this happens and no
local candidates are currently known for this same stream, the
new remote candidates are simply added to the list of remote
candidates.
</t>
<t>
Otherwise, the new candidates are used for forming candidate
pairs with the pool of local candidates and they are added to
the local check lists as described in
<xref target="cand-pairing"/>.
</t>
<t>
Once the remote agent has completed candidate gathering, it
will send an end-of-candidates event. Upon receiving such an
event, the local agent MUST update check list states as per
<xref target="state-updates"/>. This may lead to some check
lists being marked as Failed.
</t>
</section>
<section title='Receiving an End-Of-Candidates Notification'
anchor="end-of-candidates.recv">
<t>
When an agent receives an end-of-candidates notification
for a specific check list, they will update its state as per
<xref target="state-updates"/>. In case the list is still in
the Active state after the update, the agent will persist the
the fact that an end-of-candidates notification has been
received for and take it into account in future list updates.
</t>
</section>
<section title="Trickle ICE and Peer Reflexive Candidates">
<t>
Even though Trickle ICE does not explicitly modify the
procedures for handling peer reflexive candidates, their
processing could be impacted in implementations. With Trickle
ICE, it is possible that server reflexive candidates be
discovered as peer reflexive in cases where incoming
connectivity checks are received from these candidates before
the trickle updates that carry them.
</t>
<t>
While this would certainly increase the number of cases where
ICE processing nominates and selects candidates discovered as
peer-reflexive it does not require any change in processing.
</t>
<t>
It is also likely that, some applications would prefer not to
trickle server reflexive candidates to entities that are known
to be publicly accessible and where sending a direct STUN
binding request is likely to reach the destination faster than
the trickle update that travels through the signalling path.
</t>
</section>
<section title='Concluding ICE Processing'
anchor="concluding.ice">
<t>
This specification does not directly modify the procedures
ending ICE processing described in Section 8 of
<xref target="RFC5245"/>, and trickle ICE implementations will
follow the same rules.
</t>
<t>
</t>
</section>
<section title='Subsequent Offer/Answer Exchanges'
anchor="subsequent.oa">
<t>
Either agent MAY generate a subsequent offer at any time allowed
by <xref target="RFC3264"/>. When this happens agents will use
<xref target="RFC5245"/> semantics to determine whether or not
the new offer requires an ICE restart. If this is the case then
agents would perform trickle ICE as they would in an initial
offer/answer exchange.
</t>
<t>
The only differences between an ICE restart and a brand new
media session are that:
</t>
<t>
<list style='symbols'>
<t>
during the restart, media can continue to be sent to the
previously validated pair.
</t>
<t>
both agents are already aware whether or not their peer
supports trickle ICE, and there is no longer need for
performing half trickle or confirming support with other
mechanisms.
</t>
</list>
</t>
</section>
<section title='Interaction with ICE Lite'>
<t>
Behaviour of Trickle ICE capable ICE lite agents does not
require any particular rules other than those already defined
in this specification and <xref target="RFC5245"/>. This section
is hence added with an informational purpose only.
</t>
<t>
A Trickle ICE capable ICE Lite agent would generate offers or
answers as per <xref target="RFC5245"/>. Both will indicate
support for trickle ICE (<xref target="sdp.offer"/>) and given
that they will contain a complete set of candidates (the agent's
host candidates) these offers and answers would also be
accompanied with an end-of-candidates notification.
</t>
<t>
When performing full trickle, a full ICE implementation could
send an offer or an answer with no candidates and an IP6 ::
connection line address. After receiving an answer that
identifies the remote agent as an ICE lite implementation, the
offerer may very well choose to not send any additional
candidates. The same is also true in the case when the ICE lite
agent is making the offer and the full ICE one is answering. In
these cases the connectivity checks would be enough for the ICE
lite implementation to discover all potentially useful
candidates as peer reflexive. The following example illustrates
one such ICE session:
</t>
<figure title="Example " anchor="fig-ice-lite">
<artwork>
<![CDATA[
ICE Lite Bob
Agent
| Offer (a=ice-lite a=ice-options:trickle) |
|---------------------------------------------->|
| |no cand
| Answer (a=ice-options:trickle) |trickling
|<----------------------------------------------|
| Connectivity Checks |
|<--------------------------------------------->|
peer rflx| |
cand disco| |
| |
|<=============== MEDIA FLOWS =================>|
]]>
</artwork>
</figure>
<t>
In addition to reducing signaling traffic this approach also
removes the need to discover STUN bindings, or to make TURN or
UPnP allocations which may considerably lighten ICE processing.
</t>
</section>
<section title='Unilateral Use of Trickle ICE (Half Trickle)'
anchor="half-trickle">
<t>
In half trickle mode, the offerer sends a regular, vanilla ICE
offer, with a complete set of candidates. This ensures that the
offer can be processed by a vanilla ICE answerer and is mostly
meant for use in cases where support for trickle ICE cannot be
confirmed prior to sending a initial offer. The initial offer
indicates support for trickle ICE, so that the answerer can
respond with an incomplete set of candidates and continue
trickling the rest. Half trickle offers typically contain an
end-of-candidates indication, although this is not mandatory
because if trickle support is confirmed then the offerer can
choose to trickle additional candidates before it declares end
of trickling.
</t>
<t>
The half trickle mechanism can be used in cases where there is
no way for an agent to verify in advance whether a remote
party supports trickle ICE. Because it contains a full set of
candidates, its initial offer can thus be handled by a regular
vanilla ICE agent, while still allowing a trickle one to use
the optimisation defined in this specification. This prevents
negotiation from failing in the former case while still giving
roughly half the trickle ICE benefits in the latter (hence the
name of the mechanism).
</t>
<t>
Use of half trickle is only necessary during an initial
offer/answer exchange. Once both parties have received a
session description from their peer, they can each reliably
determine trickle ICE support and use it for all subsequent
offer/answer exchanges.
</t>
<t>
In some instances, using half trickle might bring more than
just half the improvement in terms of user experience. This
can happen when an agent starts gathering candidates upon user
interface cues that the user will soon be initiating an offer,
such as activity on a keypad or the phone going off hook. This
would mean that some or all of the candidate
gathering could be completed before the agent actually
needs to send the offer. Because that the answerer will be able
to trickle candidates, both agents will be able to start
connectivity checks and complete ICE processing earlier than
with vanilla ICE and potentially even as early as with full
trickle.
</t>
<t>
However, such anticipation is not not always possible. For
example, a multipurpose user agent or a WebRTC web page where
communication is a non-central feature (e.g., calling a support
line in case of a problem with the main features) would not
necessarily have a way of distinguishing between call
intentions and other user activity. In such cases, using full
trickle is most likely to result in an ideal user experience.
Even so, using half trickle would be an improvement over vanilla
ICE because it would improve the experience for answerers.
</t>
</section>
<section title='Example Flow'>
<t>
A typical successful trickle ICE exchange with an Offer/Answer
protocol would look this way:
</t>
<figure title="Example " anchor="fig-example">
<artwork>
<![CDATA[
Alice Bob
| Offer |
|---------------------------------------------->|
| Additional Candidates |
|---------------------------------------------->|
| |
| Answer |
|<----------------------------------------------|
| Additional Candidates |
|<----------------------------------------------|
| |
| Additional Candidates and Connectivity Checks |
|<--------------------------------------------->|
| |
|<=============== MEDIA FLOWS =================>|
]]>
</artwork>
</figure>
</section>
<section title='Security Considerations'>
<t>
This specification inherits most of its semantics from
<xref target="RFC5245"/> and as a result all security
considerations described there remain the same.
</t>
</section>
<section title='Acknowledgements'>
<t>
The authors would like to thank Bernard Aboba,
Flemming Andreasen, Rajmohan Banavi, Christer Holmberg,
Jonathan Lennox, Enrico Marocco, Pal Martinsen,
Martin Thomson, Dale R. Worley, and Brandon Williams
for their reviews and suggestions on improving this document.
</t>
</section>
</middle>
<back>
<references title='Normative References'>
<?rfc include="reference.RFC.2119"?>
<?rfc include="reference.RFC.3264"?>
<?rfc include="reference.RFC.4566"?>
<?rfc include="reference.RFC.5245"?>
</references>
<references title='Informative References'>
<?rfc include="reference.RFC.1918"?>
<?rfc include="reference.RFC.2543"?>
<?rfc include="reference.RFC.3261"?>
<?rfc include="reference.RFC.3388"?>
<?rfc include="reference.RFC.4787"?>
<?rfc include="reference.RFC.5389"?>
<?rfc include="reference.RFC.5766"?>
<?rfc include="reference.I-D.keranen-mmusic-ice-address-selection"?>
<?rfc include="reference.I-D.ietf-mmusic-trickle-ice-sip"?>
<reference anchor="XEP-0176">
<front>
<title>XEP-0176: Jingle ICE-UDP Transport Method</title>
<author initials='J.' surname='Beda' fullname='Joe Beda'>
<organization abbrev='Google'>Google</organization>
</author>
<author initials='S.' surname='Ludwig'
fullname='Scott Ludwig'>
<organization abbrev='Google'>Google</organization>
</author>
<author initials='P.' surname='Saint-Andre'
fullname='Peter Saint-Andre'>
<organization abbrev='Cisco'>Cisco</organization>
</author>
<author initials='J.' surname='Hildebrand'
fullname='Joe Hildebrand'>
<organization abbrev='Cisco'>Cisco</organization>
</author>
<author initials='S.' surname='Egan' fullname='Sean Egan'>
<organization abbrev='Google'>Google </organization>
</author>
<author initials='R.' surname='McQueen'
fullname='Robert McQueen'>
<organization abbrev='Collabora'>Collabora</organization>
</author>
<date month="June" year="2009" />
</front>
<seriesInfo name="XEP" value="XEP-0176" />
</reference>
<reference anchor="XEP-0030">
<front>
<title>XEP-0030: Service Discovery</title>
<author initials='J.' surname='Hildebrand'
fullname='Joe Hildebrand'>
<organization abbrev='Cisco'>Cisco</organization>
</author>
<author initials='P.' surname='Millard'
fullname='Peter Millard'>
</author>
<author initials='R.' surname='Eatmon'
fullname='Ryan Eatmon'>
</author>
<author initials='P.' surname='Saint-Andre'
fullname='Peter Saint-Andre'>
<organization abbrev='Cisco'>Cisco</organization>
</author>
<date month="June" year="2008" />
</front>
<seriesInfo name="XEP" value="XEP-0030" />
</reference>
<reference anchor="XEP-0278">
<front>
<title>XEP-0278: Jingle Relay Nodes</title>
<author initials='T.' surname='Camargo'
fullname='T. Camargo'>
</author>
<date month="June" year="2011" />
</front>
<seriesInfo name="XEP" value="XEP-0278" />
</reference>
</references>
<section title='Open Issues'>
<t>
At the time of writing of this document the authors have no
clear view on how and if the following list of issues should
be addressed.
</t>
<section title="MID/Stream Indices in SDP">
<t>
This specification does not currently define syntax for
candidate-to-stream bindings although it says that they should
be implemented with MID or a stream index. Yet, it is
reasonable to assume that most usages would need to do this
within the SDP and it may make sense to agree on the format.
Here's one possible way to do this:
<figure>
<artwork>
<![CDATA[
a=mid:1
a=candidate:1 1 UDP 1658497328 192.168.100.33 5000 typ host
a=candidate:2 1 UDP 1658497328 96.1.2.3 5000 typ srflx
a=mid:2
a=candidate:2 1 UDP 1658497328 96.1.2.3 5002 typ srflx
a=end-of-candidates
]]>
</artwork>
</figure>
</t>
</section>
<section title='Starting Checks'>
<t>
Normally vanilla ICE implementations would first activate a
check list, validate at least one pair in every component
and only then unfreeze all other checklists. With trickle ICE
this would be suboptimal since candidates can arrive randomly
and we would be wasting time waiting for a checklist to fill
(almost as if we were doing vanilla ICE). We need to decide if
unfreezing everything solely based on foundation is good
enough.
</t>
</section>
<section title='Checklist States'>
<t>
It's been proposed that we add a waiting-for-candidates
state (e.g., if the checklist is empty and no candidate
pairs have been sent or received yet).
</t>
</section>
<section title='Relationship to Continuous Nomination and
Passive Nomination'>
<t>
Does it make sense to tie trickle ICE more explicitly
the continuous nomination and passive nomination specs?
In particular, is address mobility a goal for the trickle
ICE specification?
</t>
</section>
<section title='ICE Restarts'>
<t>
We need to describe how trickle ICE interacts with ICE
restarts. Specifically, is it sufficient to modify the
ufrag and pwd without starting a full offer/answer
exchange, if the signaling protocol being used does
not require it or if the restarting entity does not
include a media description?
</t>
</section>
<section title='Candidate Redundancy and Priority'>
<t>
We need to clarify the relationship between RFC 5245
and trickle ICE with respect to candidate redundancy
and priority.
</t>
</section>
<section title='Make Trickle ICE SDP-Agnostic'>
<t>
Would it make sense to remove the tie to SDP in the
spec? This is similar to what's being done with the
ICEbis spec, so consistency might be desirable.
</t>
</section>
</section>
<section title='Interaction with ICE'
anchor='interaction'>
<t>
The ICE protocol was designed to be flexible enough to
would work in and adapt to as many network environments as
possible. Despite that flexibility, ICE as specified in
<xref target="RFC5245"/> does not by itself support trickle
ICE. This section describes how trickling of candidates
interacts with ICE.
</t>
<t>
<xref target="RFC5245"/> describes the conditions required to
update check lists and timer states while an ICE agent is in the
Running state. These conditions are verified upon transaction
completion and one of them stipulates that:
</t>
<t>
<list style='empty'>
<t>
If there is not a pair in the valid list for each component
of the media stream, the state of the check list is set to
Failed.
</t>
</list>
</t>
<t>
This could be a problem and cause ICE processing to fail
prematurely in a number of scenarios. Consider the following
case:
</t>
<t>
<list style='numbers'>
<t>
Alice and Bob are both located in different networks with
Network Address Translation (NAT). Alice and Bob themselves
have different address but both networks use the same
<xref target="RFC1918"/> block.
</t>
<t>
Alice sends Bob the candidate 10.0.0.10 which also happens
to correspond to an existing host on Bob's network.
</t>
<t>
Bob creates a check list consisting solely of 10.0.0.10 and
starts checks.
</t>
<t>
These checks reach the host at 10.0.0.10 in Bob's network,
which responds with an ICMP "port unreachable" error and per
<xref target="RFC5245"/> Bob marks the transaction as
Failed.
</t>
</list>
At this point the check list only contains Failed candidates and
the valid list is empty. This causes the media stream and
potentially all ICE processing to Fail.
</t>
<t>
A similar race condition would occur if the initial offer from
Alice only contains candidates that can be determined as
unreachable (per
<xref target="I-D.keranen-mmusic-ice-address-selection"/>) from
any of the candidates that Bob has gathered. This would be the
case if Bob's candidates only contain IPv4 addresses and the
first candidate that he receives from Alice is an IPv6 one.
</t>
<t>
Another potential problem could arise when a non-trickle
ICE implementation sends an offer to a trickle one. Consider the
following case:
<list style='numbers'>
<t>
Alice's client has a non-trickle ICE implementation
</t>
<t>
Bob's client has support for trickle ICE.
</t>
<t>
Alice and Bob are behind NATs with address-dependent
filtering <xref target="RFC4787"/>.
</t>
<t>
Bob has two STUN servers but one of them is currently
unreachable
</t>
</list>
</t>
<t>
After Bob's agent receives Alice's offer it would immediately
start connectivity checks. It would also start gathering
candidates, which would take long because of the unreachable
STUN server. By the time Bob's answer is ready and sent to
Alice, Bob's connectivity checks may well have failed: until
Alice gets Bob's answer, she won't be able to start connectivity
checks and punch holes in her NAT. The NAT would hence be
filtering Bob's checks as originating from an unknown endpoint.
</t>
</section>
<section title='Changes from Earlier Versions'>
<t>
Note to the RFC-Editor: please remove this section prior to
publication as an RFC.
</t>
<section title='Changes from draft-mmusic-trickle-ice-02'>
<t>
<list style='symbols'>
<t>
Addressed feedback from Rajmohan Banavi and Brandon Williams.
</t>
<t>
Clarified text about determining support and about how to
proceed if it can be determined that the answering agent
does not support trickle ICE.
</t>
<t>
Clarified text about check list and timer updates.
</t>
<t>
Clarified when it is appropriate to use half trickle or
to send no candidates in an offer or answer.
</t>
<t>
Updated the list of open issues.
</t>
</list>
</t>
</section>
<section title='Changes from draft-ivov-01 and draft-mmusic-00'>
<t>
<list style='symbols'>
<t>
Added a requirement to trickle candidates by order of
components to avoid deadlocks in the unfreezing algorithm.
</t>
<t>
Added an informative note on peer-reflexive candidates
explaining that nothing changes for them semantically but
they do become a more likely occurrence for Trickle ICE.
</t>
<t>
Limit the number of pairs to 100 to comply with 5245.
</t>
<t>
Added clarifications on the non-importance of how newly
discovered candidates are trickled/sent to the remote
party or if this is done at all.
</t>
<t>
Added transport expectations for trickled candidates
as per Dale Worley's recommendation.
</t>
</list>
</t>
</section>
<section title='Changes from draft-ivov-00'>
<t>
<list style='symbols'>
<t>
Specified that end-of-candidates is a media level
attribute which can of course appear as session level,
which is equivalent to having it appear in all m-lines.
Also made end-of-candidates optional for cases such as
aggressive nomination for controlled agents.
</t>
<t>
Added an example for ICE lite and trickle ICE to
illustrate how, when talking to an ICE lite agent doesn't
need to send or even discover any candidates.
</t>
<t>
Added an example for ICE lite and trickle ICE to
illustrate how, when talking to an ICE lite agent doesn't
need to send or even discover any candidates.
</t>
<t>
Added wording that explicitly states ICE lite agents
have to be prepared to receive no candidates over
signalling and that they should not freak out if this
happens. (Closed the corresponding open issue).
</t>
<t>
It is now mandatory to use MID when trickling candidates
and using m-line indexes is no longer allowed.
</t>
<t>
Replaced use of 0.0.0.0 to IP6 :: in order to avoid
potential issues with RFC2543 SDP libraries that interpret
0.0.0.0 as an on-hold operation. Also changed the port
number here from 1 to 9 since it already has a more
appropriate meaning. (Port change suggested by Jonathan
Lennox).
</t>
<t>
Closed the Open Issue about use about what to do with
cands received after end-of-cands. Solution: ignore, do
an ICE restart if you want to add something.
</t>
<t>
Added more terminology, including trickling, trickled
candidates, half trickle, full trickle,
</t>
<t>
Added a reference to the SIP usage for trickle ICE as
requested at the Boston interim.
</t>
</list>
</t>
</section>
<section title='Changes from draft-rescorla-01'>
<t>
<list style='symbols'>
<t>
Brought back explicit use of Offer/Answer. There are no
more attempts to try to do this in an O/A independent way.
Also removed the use of ICE Descriptions.
</t>
<t>
Added SDP specification for trickled candidates, the
trickle option and 0.0.0.0 addresses in m-lines, and
end-of-candidates.
</t>
<t>
Support and Discovery. Changed that section to be less
abstract. As discussed in IETF85, the draft now says
implementations and usages need to either determine
support in advance and directly use trickle, or do
half trickle. Removed suggestion about use of discovery in
SIP or about letting implementing protocols do what they
want.
</t>
<t>
Defined Half Trickle. Added a section that says how it
works. Mentioned that it only needs to happen in the first
o/a (not necessary in updates), and added Jonathan's
comment about how it could, in some cases, offer more than
half the improvement if you can pre-gather part or all of
your candidates before the user actually presses the call
button.
</t>
<t>
Added a short section about subsequent offer/answer
exchanges.
</t>
<t>
Added a short section about interactions with ICE Lite
implementations.
</t>
<t>
Added two new entries to the open issues section.
</t>
</list>
</t>
</section>
<section title='Changes from draft-rescorla-00'>
<t>
<list style='symbols'>
<t>
Relaxed requirements about verifying support following
a discussion on MMUSIC.
</t>
<t>
Introduced ICE descriptions in order to remove ambiguous
use of 3264 language and inappropriate references to
offers and answers.
</t>
<t>
Removed inappropriate assumption of adoption by RTCWEB
pointed out by Martin Thomson.
</t>
</list>
</t>
</section>
</section>
</back>
</rfc>
| PAFTECH AB 2003-2026 | 2026-04-24 01:21:59 |