WebRTC will require:
- signalling (easiest part)
- DTLS (integrated with SRTP)
- SRTP (integrated with DTLS)
DTLS is the most complicated thing, I don’t know if latest code in erlang can already do this, we have our own implementation of DTLS.
Our rust code just uses openssl implementation, it works, but in flussonic we have implementation in erlang. Source code for dtls is somewhere there in the internet, it is possible to find it.
2-3 months is very optimistic, you are really cool guys if you will do it in this time =)
If you refuse from adaptive bitrate control (you should postpone this thing) and other extended features, it is possible to do it before you will be pissed off by it.
From all our protocols (rtmp, rtsp, hls, dash, mss, mpegts, h323, sip), webrtc was most painful to implement.