--On 18 March 2004 09:24 +0100 Christian Stredicke <[EMAIL PROTECTED]> wrote:
One way with the currently existing standards is to run the call trough a trusted UA of the billing entity. That entity can measure on the RTP (or RTCP) level if the connection is still alive. Typically, this UA will be a PSTN gateway or a B2BUA (back-to-back user agent).
Some media gateways (e.g. Cisco) have RTP media inactivity timers and will tear down a SIP session if RTP goes dead (which is generally a good idea). This (IMHO) all but solves the PSTN problem.
The SIP<->SIP problem is harder. But again, if you are running an RTP proxy, you could detect inactivity there and tear down the SIP session. This would also not be a bad idea for tearing down failed or one/way connections through NATs/firewalls now I come to think of it. Clearly that's only going to work for an RTP payload.
Beyond that, it would require a change to, or some innovative (ab)use of the SIP standard to effect some form of keepalive. But note checking SIP server still has IP connectivity to the UA does not check that a functional call is still in progress. It could be (for instance) that the remote end has disconnected, the RTP's dropped etc.
Alex _______________________________________________ Sip-implementors mailing list [EMAIL PROTECTED] http://lists.cs.columbia.edu/mailman/listinfo/sip-implementors
