SIP Fundamentals

RTP and RTCP

RTP (RFC 3550) carries the actual audio/video payload between endpoints. RTCP is the companion protocol that reports stats (packet loss, jitter, round-trip time) on the same media stream. Both run on UDP, traditionally on ports 16384-32767.

RTP header anatomy

• Version: always 2.
• Payload Type (PT): 7 bits indicating codec (0=PCMU, 8=PCMA, 9=G722, 101=DTMF, etc.).
• Sequence number: increments by 1 per packet — lets the receiver detect loss.
• Timestamp: media-clock units (8000 for narrowband audio).
• SSRC: source identifier (random 32-bit per stream).
• Payload: 20ms of audio (for G.711, 160 bytes).

RTCP

RTCP runs on RTP_port+1 (or muxed via a=rtcp-mux). It carries:

• SR (Sender Report): packet count, byte count, NTP timestamp.
• RR (Receiver Report): jitter, packet loss, round-trip time.
• BYE: source leaves the session.

Use RTCP RR to diagnose call quality. If jitter > 30ms or loss > 2%, audio quality degrades.

SRTP

SRTP (RFC 3711) is RTP encrypted with AES-CTR and authenticated with HMAC-SHA1. Keys come from DTLS-SRTP (modern) or SDES (legacy). Visible on the wire as UDP/TLS/RTP/SAVP in SDP.

Related terms