This protocol operates at the Application Layer of the TCP/IP model and is often referred to as an Application Layer Framing Protocol. The RTP protocol provides a header with administrative information for packetized real-time media such as voice and video.
The media samples are prepared for transport and RTP together with its companion protocol RTCP (Real-Time Transport Control Protocol), attempt to provide reliable delivery of the media packets. Within the RTP header is a method of unique identification using an SSRC (Synchronising Source ID), sequence numbers for reliable delivery, payload identification and finally timestamping.
The sequence numbers allow the receiving equipment to reconstruct the packet stream in the correct order, while the timestamp allows the receiver to buffer the data in a de-jitter buffer to smooth out jitter. The Synchronising Source ID is used to identify the sending station and the codec in use is indicated in the payload type field.
A sister protocol, RTCP uses a series of Sender and Receiver reports to monitor the flow of data and indicate packet loss and jitter.
When the media samples are encapsulated within an RTP packet, this in turn is encapsulated inside a UDP datagram and finally inside an IP packet. The RTP, UDP and IP headers produce a total of 40 bytes of header information or overhead as it is often known. Only a small amount of media, typically 20 bytes is encapsulated in each RTP packet which means that as much as two thirds of bandwidth required to transmit the packet stream is being used to forward overhead. For this reason, particularly on slow links, header compression is often used to reduce the amount of bandwidth used for transmission of the stream. Alternatively, adding additional media to RTP packets will reduce the ratio of overhead to media.
Within some of the network equipment such as endpoints, gateways or routers, RTP translators are sometimes used to translate from one codec to another to accommodate a particular network requirement. For example, if a number of 1.5Mbps MPEG video streams are in a multimedia multicast conference and these streams have to traverse a 1Mbps circuit, a translator could be employed to translate or transcode the 1.5Mbps video streams to another format such as H.261 producing 3 x 256Kbps streams that could be passed over the 1Mps circuit. would swamp the 1Mbps circuit so a Translator is employed to further compress the video into 3 x 256Kbps streams using the ITU-T H.261 video compression standard.
RTP and RTCP is covered in detail on some the training courses run by NSTUK Ltd such as the 2 day Introduction to Voice over IP and the 3 day VoIP with SIP course.
Take a look at the NSTUK Training Schedule for information on scheduled training courses or enquire about an On-Site training course.