RTP: definition and a practical example
Short answer: RTP (Real-time Transport Protocol), defined in RFC 3550, is the protocol that carries a VoIP call's audio in numbered, timestamped packets.
In Portuguese: RTP
What is RTP?
RTP (Real-time Transport Protocol) is the protocol that carries the voice on a VoIP call. While SIP sets up the call, RTP carries the audio itself, in small packets sent at a fixed pace, usually one every 20 ms. It is defined in RFC 3550 and runs over UDP.
Each RTP packet has a 12-byte header with four main pieces of information: the payload type, which identifies the codec; the sequence number, which makes lost or out-of-order packets visible; the timestamp, used to play the audio at the right pace and to calculate jitter; and the SSRC, which identifies the source of the stream. Each direction of the call is a separate stream, with its own ports, agreed in the SDP.
RTCP
RTCP (RTP Control Protocol), defined in the same RFC 3550, travels alongside RTP and carries no voice. It sends periodic reports with packets sent, fraction lost and interarrival jitter. Traditionally it uses the next odd port after RTP’s.
SRTP
SRTP (Secure RTP), from RFC 3711, encrypts the audio payload and authenticates the packets. Anyone who captures the traffic sees the headers but cannot hear the conversation.
In practice
A test call of about 7 seconds with 20 ms packets generates some 350 packets per direction. A log line such as “RTP 343 sent / 340 received” shows:
| Measure | Value |
|---|---|
| Packets sent | 343 |
| Packets received | 340 |
| Lost | 3 (0.9%) |
| Audio lost | about 60 ms |
If the received side showed 0, the call would have been answered with no audio in that direction, a typical NAT or SIP ALG case.
How Monitor uses this
The evidence for each incident shows the RTP packets sent and received on each test call, for example “RTP 343 sent / 340 received”. Monitor checks that audio flows in both directions, so a call that is answered but silent on one side does not pass as healthy. See voice quality with MOS-LQO.
Frequently asked questions
Why does the call connect but have no audio?
Because SIP and RTP take different paths. SIP can set up the call normally while the RTP packets are blocked by a firewall, sent to a private IP because of NAT or rewritten by a SIP ALG. The result is an answered call that is silent in both directions or in just one.
What is the difference between RTP and SRTP?
SRTP (RFC 3711) is RTP with encryption, authentication and replay protection. The packet format and pace are the same; the difference is that the audio payload travels encrypted and cannot be heard by anyone who captures the traffic. The keys are agreed in the call signaling.