&xep0167; recommends the use of the Secure Real-time Transport Protocol (SRTP) for end-to-end encryption of RTP sessions negotiated using &xep0166;. An alternative approach to end-to-end encryption of RTP traffic is provided by &rfc6189;, developed by Phil Zimmermann, the inventor of "Pretty Good Privacy" (PGP). Although negotiation of ZRTP mainly occurs in the media channel rather than the signalling channel, the ZRTP specification defines one SDP attribute called "zrtp-hash" (this communicates the ZRTP version supported as well as a hash of the Hello message). Inclusion of this information is OPTIONAL in both SIP/SDP and Jingle.

The SDP format is shown below.

An example follows.

This SDP attribute can be translated into Jingle as a <zrtp-hash/> element qualified by the 'urn:xmpp:jingle:apps:rtp:zrtp:1' namespace, as shown below.

An example follows.

The <zrtp-hash/> element is sent as a child of the <encryption/> element defined in &xep0167;.

If the Jingle initiator wishes to use ZRTP, it includes the <zrtp-hash/> element in its session invitation (where it hashes over its own Hello message as described in the ZRTP specification).

If the receiving party wishes to proceed with ZRTP negotiation, it also includes the <zrtp-hash/> element in its session-accept message (where it hashes over its own Hello message as described in the ZRTP specification).

Note that a unique zrtp-hash is needed for each media stream, since the hash for each stream is computed from a different ZRTP Hello message (e.g., if a session includes both audio and video then the value of the <zrtp-hash/> element included in the <description/> element for the audio stream will be different from the value for the video stream).

If an entity supports the zrtp-hash session-info message, it MUST advertise that fact in its responses to &xep0030; information ("disco#info") requests by returning a feature of "urn:xmpp:jingle:apps:rtp:zrtp:1":

In order for an application to determine whether an entity supports this protocol, where possible it SHOULD use the dynamic, presence-based profile of service discovery defined in &xep0115;. However, if an application has not received entity capabilities information from an entity, it SHOULD use explicit service discovery instead.

Security considerations for ZRTP itself are provided in RFC 6189.

XMPP stanzas such as Jingle session-info messages and service discovery exchanges are not encrypted or signed. As a result, it is possible for an attacker to intercept these stanzas and modify them, thus convincing one party that the other party does not support ZRTP and therefore denying the parties an opportunity to use ZRTP. However, because the zrtp-hash is mostly advisory, the parties could still use ZRTP even if the signalling channel is compromised.

This document requires no interaction with &IANA;.

Thanks to Werner Dittmann and Emil Ivov for their implementation feedback.