T.140 Real-time Text Conversation over WebRTC Data Channels

The ITU-T Protocol for multimedia application text conversation (Recommendation ITU-T T.140) defines a protocol for text conversation, also known as real-time text. The native transport for IP networks is the "RTP Payload for Text Conversation" mechanism, based on the Real-time Transport Protocol (RTP) . This document specifies how a WebRTC data channel can be used as a transport mechanism for T.140, and how the SDP offer/answer mechanism can be used to negotiate such data channel. In this document, a T.140 data channel refers to a WebRTC data channel for which the instantiated sub-protocol is "t140", and where the channel is negotiated using the SDP-based external negotiation method . NOTE 1: This WebRTC term of a "T.140 data channel" is actually synonym to the originally introduced concept of a "T.140 data channel" for the T.140 protocol, see Section 4.3 of . NOTE 2: The decision to transport realtime text over a data channel, instead of using RTP based transport , in WebRTC is constituted by use-case "U-C 5: Realtime text chat during an audio and/or video call with an individual or with multiple people in a conference", see Section 3.2 of . The brief notation "T.140" is used as a synonym for the text conversation protocol according to . defines the generic data channel properties for a T.140 data channel, and defines how they are conveyed in an SDP dcmap attribute. While this document defines how to establish a T.140 data channel using the SDP-based external negotiation method , the generic T.140 and gateway considerations defined in , and of this document can also be applied when a T.140 data channel is established using another mechanism (e.g., the mechanism defined in ). Section 5 of defines the mapping between the SDP dcmap attribute parameters and the protocol parameters used in . This document is based on an earlier Internet draft edited by Keith Drage, Juergen Stoetzer-Bradler and Albrecht Schwarz.

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 when, and only when, they appear in all capitals, as shown here.

The following WebRTC data channel property values apply to a T.140 data channel: Subprotocol Identifier t140 Transmission reliability reliable Transmission order in-order Label See and

The generic SDP considerations, including the SDP Offer/Answer procedures, for negotiating a WebRTC data channel are defined in . This section defines the SDP considerations that are specific to a T.140 data channel.

An offerer and answerer MUST, in each offer and answer, include an SDP 'dcmap' attribute in the SDP media descripton (m= section) describing the SCTP association used to realize the T.140 data channel. The offerer and answerer MUST include the subprotocol attribute parameter, with a "t140" parameter value, in the 'dcmap' attribute value. The offerer and answerer MAY include the priority attribute parameter and the label attribute parameter in the 'dcmap' attribute value. If the offerer includes a label attribute parameter, the answerer MUST NOT change the value in the answer. NOTE: As specified in , when a data channel is negotied using the mechanism defined in , the label attribute parameter value has to be the same in both directions. That rule also applies to data channels negotiated using the mechanism defined in this document. The offerer and answerer MUST NOT include the max-retr and max-time attribute parameters in the 'dcmap' attribute. If the ordered attribute parameter is included in the 'dcmap' attribute, it MUST be assigned the value 'true'. Below is an example of the 'dcmap' attribute for a T.140 data channel with stream id=3 and without any label: a=dcmap:3 subprotocol="t140"

An offerer and answerer MAY, in each offer and answer, include an SDP 'dcsa' attribute in the m= section describing the SCTP association used to realize the T.140 data channel. If an offerer or answere receives a 'dcsa' attribute that contains an SDP attribute which usage has not been defined for a T.140 data channel, the offerer or answerer should ignore the 'dcsa' attribute, following the rules in . NOTE: At the time of writing this specification, the latest verion of the API defined in does not support the use of the SDP attributes defined in this section.

A 'dcsa' attribute can contain the SDP 'fmtp' attribute used to indicate a maximum character transmission rate . The 'cps' attribute parameter is used to indicate the maximum character transmission rate that the endpoint that includes the attribute is able to receive, and the value is used as a mean value in characters per second over any 10-second interval. If the 'fmtp' attribute is included, the 'format' attribute parameter MUST be set to "-". If the 'fmtp' attribute is not included, it indicates that no maximum character transmission rate is indicated. It does not mean that the default value of 30 applies . The offerer and answerer MAY modify the 'cps' attribute parameter value in subsequent offers and answers. This document does not define any other usage of the 'fmtp' attribute for a T.140 channel. If an offerer or answerer receives a 'dcsa' attribute that contains 'fmtp' attribute that is not according to the procedure above the offerer or answerer MUST discard the 'dcsa' attribute. NOTE: The 'cps' attribute parameter is especially useful when a T.140 data channel endpoint is acting as a gateway [] and is interworking with a T.140 transport mechanism that have restrictions on how many characters can be sent per second.

'dcsa' attributes can contain the SDP 'hlang-send' and 'hlang-recv' attributes to negotiate the language to be used for the real-time text conversation. For a T.140 data channel, the modality is "written" .

'dcsa' attributes can contain the SDP 'sendonly', 'recvonly', 'sendrecv' and 'inactive' attributes to negotite the direction in which text can be transmitted in a real-time text conversation. NOTE: A WebRTC data channel is always bi-directional. The usage of the 'dcsa' attribute only affects the direction in which implementations are allowed to transmit text on a T.140 data channel. The offer/answer rules for the direction attributes are based on the rules for unicast streams defined in , as described below. Note that the rules only apply to the direction attributes. Session level direction attributes have no impact on a T.140 data channel. An offerer and answerer MUST mark the direction of the text by sending a direction attribute inside 'dcsa' attribute.

If the offerer wishes to both send or receive text on a T.140 data channel, it SHOULD mark the data channel as sendrecv with a 'sendrecv' attribute inside a 'dcsa' attribute. If the offerer does not explicitly mark the data channel, a 'sendrecv' attribute inside a 'dcsa' attribute is implicitly applied. If the offerer wishes to only send text on a T.140 data channel, it MUST mark the data channel as sendonly with a 'sendonly' attribute inside a 'dcsa' attribute. If the offerer wishes to only receive text on a T.140 data channel, it MUST mark the data channel as recvonly with a 'recvonly' attribute inside a 'dcsa' attribute. If the offerer wishes to neither send or receive text on a T.140 data channel, it MUST mark the data channel as inactive with a 'inactive' attribute inside a 'dcsa' attribute. If the offerer has marked a data channel as sendrecv (implicit or explicit) or recvonly, it MUST be prepared to receive T.140 data as soon as the state of the T.140 data channel allows it. When the offerer receives an answer to the offer, if the answerer has marked a data channel as sendrecv (implicit or explicit) or recvonly in the answer, the offerer can start to send T.140 data as soon as the state of the T.140 data channel allows it. If the answerer has marked the data channel as inactive or sendonly, the offerer MUST NOT send any T.140 data. As the answerer implementation might not support the procedures in this section, if the answerer has not marked the direction of a T.140 data channel in accordance with the procedures above, it is RECOMMENDED that the offerer does not process that as an error situation, but rather assume that the answerer might both send and receive T.140 data on the data channel.

When the answerer accepts an offer, and marks the direction of the text in the corresponding answer, the marking is based on the marking (explicit or implicit) in the offer. If the offerer marked the data channel as sendrecv (implicitly or explicitly), the answerer MUST mark the data channel as sendrecv, sendonly, recvonly or inactive with a 'sendrecv', 'sendonly', 'recvonly' respective 'inactive' attribute inside a 'dcsa' attribute. If the answerer does not explicitly mark the data channel, a 'sendrecv' attribute inside a 'dcsa' attribute is implicitly applied. If the offerer marked the data channel as sendonly, the answerer MUST mark the data channel as recvonly or inactive with a 'recvonly' respective 'inactive' attribute inside a 'dcsa' attribute. If the offerer marked the data channel as recvonly, the answerer MUST mark the data channel as sendonly or inactive with a 'sendonly' respective 'inactive' attribute inside a 'dcsa' attribute. If the offerer marked the data channel as inactive, the answerer MUST mark the data channel as inactive with a 'inactive' attribute inside a 'dcsa' attribute. If the answerer has marked a data channel as sendrecv or recvonly, it MUST be prepared to receive data as soon as the state of the T.140 data channel allows transmission of data.

If an endpoint wishes to modify a previously negotiated text direction in an ongoing session, it MUST initiate an offer that indicates the new direction, following the rules in . If the answerer accepts the offer it follows the procedures in .

Below is an example of an m= section of an offer for a T.140 data channel offering real-time text conversation in Spanish and Esperanto, and an m= section in the associated answer accepting Esperanto. The maximum character transmission rate is set to 20, and the default text transmission direction "sendrecv", apply.

Below is an example of an m= section of an offer for a T.140 data channel where the offerer wishes to only receive real-time text, and an m= section in the associated answer indicating that the answerer will only send real-time text. The default maximum character transmission rate 30 applies. No language to be used for the real-time text conversation is indicated.

Section 6.1 of describes the generic T.140 session control functions at high-level and a signalling protocol independent manner. The list below describes how the functions are realized when using a T.140 data channel. Prepare session: An endpoint can indicate its support of T.140 data channels using signalling specific means (e.g., using SIP OPTIONS ), or by indicating the support in an offer or answer () Initiate session: An offer used to request the establishment of a T.140 data channel () Accept session: An answer used to accept a request to establish a T.140 data channel () Deny session: An answer used to reject a request to establish a T.140 data channel, using the generic procedures for rejecting a data channel Disconnect session: An offer or answer used to disable a previously established T.140 data channel, using the generic procedures for closing a data channel Data: Data sent on an established T.140 data channel ()

T.140 text is encoded and framed as T140blocks . Each T140block is sent on the SCTP stream used to realize the T.140 data channel using standard T.140 transmission procedures . One or more T140blocks can be sent in a single SCTP user message . Unlike RTP based transport for realtime text , T.140 data channels do not use redundant transmission of text. The reason for this is that the T.140 data channel achieves robust transmission by using the "reliable" mode of the data channel. Data sending and reporting procedures conform to . See Section 8 of for coding details.

As described in , buffering can be used to reduce overhead, with the maximum buffering time being 500 ms. It can also be used for staying within the maximum character transmission rate (), if such has been provided by the peer. An implementation needs to take the user requirements for smooth flow and low latency in real-time text conversation into consideration when assigning a buffer time. It is RECOMMENDED to use the default transmission interval of 300 milliseconds , or lower, for T.140 data channels.

In case of network failure or congestion, T.140 data channels might fail and get torn down. If this happens but the session sustains, it is RECOMMENDED that a low number of retries are made to reestablish the T.140 data channels. If reestablishment of the T.140 data channel is successful, an implementation MUST evaluate if any T140blocks were lost. Retransmission of already successfully transmitted T140blocks MUST be avoided, and missing text markers SHOULD be inserted in the received data stream where loss is detected or suspected. NOTE: If the the SCTP association used to realize the T.140 data channel fails and get torn down, it needs to be re-established before the T.140 data channel can be reestablished. The procedure after the reestablishment of the T.140 data channel defined in this section apply no matter if only the T.140 data channel, or the whole SCTP association, got torn down.

If an implementation needs to support multi-party scenarios, the implementation needs to support multiple simultaneous T.140 data channels, one for each remote party. At the time of writing this document, this is true even in scenarios where each participant communicate via a centralized conference server. The reason is that, unlike RTP media, WebRTC data channels and the T.140 protocol do not support the indication of the source of T.140 data. The SDP 'dcmap' attribute label attribute parameter () can be used by the offerer to provide additional information about each T.140 data channel, and help implementations to distinguish between them. NOTE: Future extensions to T.140, or to the T140block, might allow indicating the source of T.140 data, in which case it might be possible to use a single T.140 data channel to transport data from multiple remote sources. That is useful for multi-party aware clients that are able present the conference in a way that is adapted to user expectations regarding presentation style and real-time performance. Conference mixers that use a single T.140 data channel to transmit real-time text towards clients might, without any protocol extensions, produce a multi-party presentation completely within the text stream, and with limitations in real-time performance and presentation style.

A number of real-time text transports and protocols have been defined for both packet switched and circuit switched networks. Many are based on the ITU-T T.140 protocol on application and presentation level . At the time of writing this document, some mechanisms are no longer used, as the technologies they use have been obsoleted etc, while others are still in use. When performing interworking between T.140 data channels and real-time text in other transports and protocols, a number of factors need to be considered. At the time of writing this document, the most common IP-based real-time text transport is the RTP based mechanism defined in . While this document does not define a complete interworking solution, this list below provides some guidance and considerations to take into account when designing a gateway for interworking between T.140 data channels and RTP-based T.140 transport: For each T.140 data channel there is an RTP stream for real-time text . Redundancy is by default declared and used on RTP stream. On the T.140 data channel there is no redundancy, but the reliable property of T.140 the data channel is set. During a normal text flow, T140blocks received from one network are forwarded towards the other network. Keep-alive traffic is implicit on the T.140 data channel. A gateway might have to extract keep-alives from incoming RTP streams, and MAY generate keep-alives on outgoing RTP streams. If the gateway detects or suspects loss of data on the RTP stream, the gateway gateway SHOULD insert the T.140 missing text marker in the data sent on the outgoing T.140 data channel. If the gateway detects that the T.140 data channel has failed and got torn down, once the data channel has been reestablished the gateway SHOULD insert the T.140 missing text marker in the data sent on the outgoing RTP stream if it detects or suspects that data on the T.140 data channel was lost. The gateway MUST indicate the same text transmission direction () on the T.140 data channel and the RTP stream. NOTE: In order for the gateway to insert a missing text marker, or to perform other actions that require that the gateway has access to the T.140 data, the T.140 data cannot be encrypted end-to-end between the T.140 data channel endpoint and the RTP endpoint. At the time of writing this document, a mechanism to provide such end-to-end encryption has not been defined.

This document updates RFC 8373, by defining how the SDP hlang-send and hlang-recv attributes are used for the "application/webrtc-datachannel" media type. SDP offerers and answerers MUST NOT include the attributes directly in the m= section associated with the 'application/webrtc-datachannel' media type. Instead, the attributes MUST be associated with individual data channels, using the SDP 'dcsa' attribute. A specification that defines a subprotocol that uses the attributes MUST specify the modality for that subprotocol, or how to retreive the modality if the subprotocol supports multiple modalities.

The generic security considerations for WebRTC data channels are defined in . As data channels are always encrypted by design, the T.140 data channels will also be encrypted. The generic security considerations for the SDP-based external negotiation method are defined in .

[RFC EDITOR NOTE: Please replace all instances of RFCXXXX with the RFC number of this document.]

This document adds the subprotocol identifier "t140" to the "WebSocket Subprotocol Name Registry" as follows: Subprotocol Identifier: t140 Subprotocol Common Name: ITU-T T.140 Subprotocol Definition: RFCXXXX Reference: RFCXXXX

This document modifies the usage of the SDP 'fmtp' attribute, if this attribute is included in an SDP 'dcsa' attribute and associated with an T.140 real-time text session over a WebRTC data channel. The modified usage is described in . The usage level "dcsa(t140)" is added to the IANA registration of the SDP 'fmtp' attribute as follows: Contact name: MMUSIC Chairs Contact email: mmusic-chairs@ietf.org Attribute name: fmtp Usage level: dcsa(t140) Purpose: Indicate the maximum transmission rate that an endpoint is willing to recive on a T.140 data channel. Reference: RFCXXXX

This document modifies the usage of the SDP 'hlang-send' and 'hlang-recv' attributes, if these attributes are included in SDP 'dcsa' attributes associated with an T.140 data channel. The modified usage is described in . The usage level "dcsa(t140)" is added to the IANA registration of the SDP 'hlang-send' attribute as follows: Contact name: MMUSIC Chairs Contact email: mmusic-chairs@ietf.org Attribute name: hlang-send Usage level: dcsa(t140) Purpose: Negotiate the language to be used on a T.140 data channel. Reference: RFCXXXX The usage level "dcsa(t140)" is added to the IANA registration of the SDP 'hlang-recv' attribute as follows: Contact name: MMUSIC Chairs Contact email: mmusic-chairs@ietf.org Attribute name: hlang-recv Usage level: dcsa(t140) Purpose: Negotiate the language to be used on a T.140 data channel. Reference: RFCXXXX

This document modifies the usage of the SDP 'sendonly', 'recvonly', 'sendrecv' and 'inactive' attributes, if these attributes are included in SDP 'dcsa' attributes associated T.140 data channel. The modified usage is described in . The usage level "dcsa(t140)" is added to the IANA registration of the SDP 'sendonly' attribute as follows: Contact name: MMUSIC Chairs Contact email: mmusic-chairs@ietf.org Attribute name: sendonly Usage level: dcsa(t140) Purpose: Negotiate the direction in which real-time text can be sent on a T.140 data channel. Reference: RFCXXXX The usage level "dcsa(t140)" is added to the IANA registration of the SDP 'recvonly' attribute as follows: Contact name: MMUSIC Chairs Contact email: mmusic-chairs@ietf.org Attribute name: recvonly Usage level: dcsa(t140) Purpose: Negotiate the direction in which real-time text can be sent on a T.140 data channel. Reference: RFCXXXX The usage level "dcsa(t140)" is added to the IANA registration of the SDP 'sendrecv' attribute as follows: Contact name: MMUSIC Chairs Contact email: mmusic-chairs@ietf.org Attribute name: sendrecv Usage level: dcsa(t140) Purpose: Negotiate the direction in which real-time text can be sent on a T.140 data channel. Reference: RFCXXXX The usage level "dcsa(t140)" is added to the IANA registration of the SDP 'inactive' attribute as follows: Contact name: MMUSIC Chairs Contact email: mmusic-chairs@ietf.org Attribute name: inactive Usage level: dcsa(t140) Purpose: Negotiate the direction in which real-time text can be sent on a T.140 data channel. Reference: RFCXXXX

This document is based on an earlier Internet draft edited by Keith Drage, Juergen Stoetzer-Bradler and Albrecht Schwarz. Thomas Belling provided useful comments on the initial (pre-submission) version of the draft. Gunnar Hellstrom provided comments and text on the draft. Paul Kyzivat provides comments on the draft.