Re: [rtcweb] Interest and need for Websocket subprotocol - JSEP over websockets

[[e-mail]]

Hello all

While I agree SIP over Websockets is default signaling protocol for 
WebRTC while working with IMS, there could be scenarios where WebRTC 
calls can get initiated from non SIP UAs like web browsers which do not 
support SIP. Then in such cases, the following things could happen
1) the WebRTC client on the browser can use JSEP to send its signaling 
information over WebSocket,
2) the JSEP message would then land on the WebRTC GW over WS.
3) This JSEP message would then be converted to a SIP message and then 
sent to IMS core.
4) within IMS core, its a regular SIP message
5) Again in the reverse direction, WebRTC GW would convert SIP to JSEP
6) JSEP message is sent over Websocket to UE.

now we see JSEP messages getting exchanged over Websockets. so if the 
websocket sub-protocol does not define the type as "jsep", then the 
WebRTC GW would not know the incoming message type and hence may discard 
it or its behavior may be uncertain.

Also the JSEP message needs to be enhanced to add more message types 
(along with current OFFER / ANSWER) to be able to map it with standard 
signaling protocol like SIP as defined in 
https://tools.ietf.org/html/draft-partha-rtcweb-jsep-sip-01

Regards
Ranjit

On 2014-12-03 12:40 pm, Makaraju, Maridi Raju (Raju) wrote:
> + 1 for using SIP over WebSocket.
> 
> FROM: rtcweb [mailto:[e-mail]] ON BEHALF OF Roman
> Shpount
>  SENT: Wednesday, December 03, 2014 12:38 PM
>  TO: [e-mail]
>  CC: [e-mail]
>  SUBJECT: Re: [rtcweb] Interest and need for Websocket subprotocol -
> JSEP over websockets
> 
> Is there any reason you cannot use SIP over WebSocket
> (https://tools.ietf.org/html/rfc7118 [1])?
> 
> Call signaling will require a lot more information then what is
> provided in JSEP. JSEP mostly deals with offer and answer processing.
> Signaling will also need to deal with things like who is calling, why
> they are calling, transfers, other application specific details. In
> other words, I think this is a very bad idea.
> 
> _____________
>  Roman Shpount
> 
> On Wed, Dec 3, 2014 at 1:31 PM, <[e-mail]> wrote:
> 
> Hi
>  With websockets as a de-facto transport protocol for WebRTC signaling
> and JSEP being the format of encoding information, there is a need for
> a defining a websocket sub-protocol : jsep. So I would like to know if
> there is any interest in the community and also the views from experts
> about the need for a websocket-sub protocol for JSEP.
> 
>  The main purpose of defining the sub protocol (jsep) is to make sure
> that the WebRTC client (WIC) and WebRTC server (E-CSCF) are receiving
> JSEP encoded messages.
> 
>  Thanks
>  Ranjit
> 
>  _______________________________________________
>  rtcweb mailing list
>  [e-mail]
>  https://www.ietf.org/mailman/listinfo/rtcweb [2]
> 
> 
> 
> Links:
> ------
> [1] https://tools.ietf.org/html/rfc7118
> [2] https://www.ietf.org/mailman/listinfo/rtcweb