What is the feasibility of using OONI to test for blocking of WebRTC?
When the Snowflake pluggable transport is deployed, it will be good to know when and how WebRTC connections are blocked.
There are some complications. WebRTC is not one protocol but rather a suite of protocols including ICE, DTLS, and SRTP. Different forms of WebRTC will have different fingerprints.
What would be a good basic test of WebRTC availability? Perhaps a simulated connection to Google Hangouts.
Maybe not google hangouts, since it suffers from the 'google is blocked in china' phenomenon.
Getting a full WebRTC client will remain problematic - the libRTC library used by both snowflake and chrome / firefox is a huge hassle to link and include in a process - especially given the mobile considerations.
A good start might be to validate STUN and TURN functionality - and there are much smaller libraries for both of those protocols.
The other test worth having in place is seeing if a DTLS connection can be established - probably to a backend infrastructure node participating in the experiment.
--Will
On Fri, Mar 10, 2017 at 12:29:50PM -0800, David Fifield wrote:
What is the feasibility of using OONI to test for blocking of WebRTC?
When the Snowflake pluggable transport is deployed, it will be good to know when and how WebRTC connections are blocked.
There are some complications. WebRTC is not one protocol but rather a suite of protocols including ICE, DTLS, and SRTP. Different forms of WebRTC will have different fingerprints.
What would be a good basic test of WebRTC availability? Perhaps a simulated connection to Google Hangouts. _______________________________________________ ooni-dev mailing list ooni-dev@lists.torproject.org https://lists.torproject.org/cgi-bin/mailman/listinfo/ooni-dev
On Fri, Mar 10, 2017 at 12:41:11PM -0800, Will Scott wrote:
A good start might be to validate STUN and TURN functionality - and there are much smaller libraries for both of those protocols.
The other test worth having in place is seeing if a DTLS connection can be established - probably to a backend infrastructure node participating in the experiment.
I agree, STUN and TURN are probably the best place to start.
I did a cursory search to see if there are any public DTLS servers (thinking that OONI could test connections to them), but didn't find any. However I did find that some Internet-of-things standards (LWM2M and oneM2M) use DTLS. https://en.wikipedia.org/wiki/OMA_LWM2M
-
David Fifield -
Will Scott