next up previous contents index
Next: Challenges of Multimedia Transmission Up: State of the Art Previous: Other Works in Objective   Contents   Index


Multimedia Transport Protocols

The goal of multimedia transport protocols is to transmit multimedia signals from one point to another point. These points are connected by communication network employing specific protocols. Generally, multimedia original signals are encoded to reduce the bit rate. When the encoded stream is to be sent to another location in the network, the transport protocols are responsible for the packetization and the delivery of that stream. At the other side, the encoded multimedia stream is reconstructed from the stream of delivered packets and then decoded to produce a useful multimedia signal to be played back or stored for further use. The Internet Protocol (IP) is a packet-based network protocol used to exchange data over networks. It is the underlying network protocol, i.e. other protocols are built over IP. The most used higher level protocol is the Transport Control Protocol (TCP), which is a reliable transport protocol designed for data transmission and extensively used in Internet services. TCP is not suitable for real-time applications as the retransmissions can lead to high delay and cause delay jitter, which significantly degrades the quality. In addition, it does not support multicast. Also, congestion control mechanisms, namely slow start, are not suitable for audio or video media transmission. On the other hand the transport protocol that is generally used for real-time multimedia transmission is the User Datagram Protocol (UDP). UDP does not guarantee the arrival of the packet, it is up to the application or higher level protocols to take care of the sent data. The most used, for real time applications, protocol which is built over UDP is Real time Transport Protocols (RTP). The most important variables governing the operation of RTP are the Time Stamp (TS) and the Sequence Number (SN). The TS is responsible for placing the incoming packets in correct timing order. The initial value of the TS is selected randomly and independently for each RTP stream. The TS value is increased by the time indicated by each packet. For example, for the case of audio transmission with 20 ms as packetization interval, the TS may take the values 0, 20, 40, 60, ...for packet numbers 1, 2, 3, ... respectively. The SN is used to detect packet loss occurrences. It is increased by one for each packet in the stream. It should be mentioned that for a video frame that is split into multiple RTP packets, these packets share the same value of TS but use different SN. There is a separate control protocol that is generally used with RTP, which is named Real Time Control Protocol (RTCP). RTCP synchronizes across different media streams by feedback messages (e.g. the sender report). It also provides feedback on the quality of data transmission by using lost packet counts in the Receiver Report. In addition, it identifies and keeps track of the participants. RTCP reports are sent periodically (every 5 sec.) between participants with the restriction that its traffic should not exceed 5% of the total data traffic. RTP supports multicasting, payload type identification, time stamping, sequence numbering, delivery monitoring. In addition, the underlying UDP protocol supports multiplexing and checksum services. Even if RTP is the most used protocol for real-time applications, it has some problems. First, it does not support the timely delivery of data or any other QoS guarantee. In-time delivery requires lower layers that have control over resources in switches or routers (e.g. Resource Reservation Protocol, RSVP). Second, it does not guarantee delivery, so packets may be delivered out of order or get lost. In addition, there is no mechanism to recover from packet loss.

Subsections
next up previous contents index
Next: Challenges of Multimedia Transmission Up: State of the Art Previous: Other Works in Objective   Contents   Index
Samir Mohamed 2003-01-08