University of Delaware Real-Time Protocols RTP/RTCP/RTSP Amit Hetawal University of Delaware CISC 856 -Fall 2005 Thanks to Professor Amer

Overview History of streaming media Streaming performance requirements Protocol stack for multimedia services Real-time transport protocol (RTP) RTP control protocol (RTCP) Real-time streaming protocol (RTSP)

Brief history of streaming media

Real-time multimedia streaming Real-time multimedia applications Video teleconferencing Internet Telephony (VoIP) Internet audio, video streaming (A-PDUs)

Streaming performance requirements Sequencing – to report PDU loss to report PDU reordering to perform out-of-order decoding Time stamping and Buffering for play out for jitter and delay calculation Payload type identification for media interpretation Error concealment –covers up errors from lost PDU by using redundancy in most-adjacent-frame Quality of Service (QoS) feedback – from receiver to sender for operation adjustment Rate control –sender reduces sending rate adaptively to network congestion

Ideal Timing – no jitter 30 seconds 00.00.00 00.00.10 First RTP-PDU application 00.00.11 00.00.20 Second RTP-PDU 00.00.21 00.00.30 Third RTP-PDU 00.00.31 Send time Play time

Reality – jitter Send time Play time delay 00.00.00 00.00.10 First RTP-PDU 00.00.11 00.00.20 Second RTP-PDU 00.00.21 00.00.30 00.00.25 00.00.40 Third RTP-PDU 00.00.35 00.00.37 Fourth RTP-PDU 00.00.41 00.00.47 Send time 00.00.51 Play time

Jitter (contd.) Send time Play time 00.00.00 00.00.10 First RTP-PDU(0) 00.00.11 00.00.20 Second RTP-PDU(10) 00.00.21 00.00.30 00.00.18 00.00.25 00.00.40 Third RTP-PDU(20) 00.00.28 00.00.35 00.00.37 Fourth RTP-PDU (30) 00.00.38 00.00.41 00.00.47 Send time 00.00.48 00.00.51 Play time 00.00.58

Jitter (contd.) Playback buffer At time 00:00:18 At time 00:00:28

How does Sequence number and Timestamp help ? Audio silence example: sender receiver Seq no.1, Tmpst 100 Seq no.2, Tmpst 200 Seq no.3, Tmpst 300 Seq no.4, Tmpst 600 Seq no.5, Tmpst 700 Consider audio data What should the sender do during silence? silence Not send anything Why might this cause problems? Receiver cannot distinguish between loss and silence Solution: After receiving no PDUs for a while, next PDU received at the receiver will reflect a big jump in timestamp, but have the correct next seq. no. Thus, receiver knows what happened.

Streaming performance requirements Sequencing – to report PDU loss to report PDU reordering to perform out-of-order decoding Time stamping and Buffering for play out for jitter and delay calculation Payload type identification for media interpretation Error concealment –covers up errors from lost PDU by using redundancy in most-adjacent-frame Quality of Service (QoS) feedback – from receiver to sender for operation adjustment Rate control –sender reduces sending rate adaptively to network congestion

Support from transport layers TCP is not used because: TCP does retransmissions  unbounded delays No provision for time stamping TCP does not support multicast TCP congestion control (slow-start) unsuitable for real-time transport RTP + UDP usually used for multimedia services

Protocol stack for multimedia services RTSP RTP RTCP TCP (till now)

RTP: Introduction Application RTP RTCP UDP IP Data Link Physical Provides end-to-end transport functions for real-time applications Supports different payload types All RTP and RTCP PDUs are sent to same multicast group (by all participants) All RTP PDUs sent to an even-numbered UDP port, 2p All RTCP PDUs sent to UDP port 2p+1 Does NOT provide timely delivery or other QoS guarantees Relies on other protocols like RTCP and lower layers Does NOT assume the underlying network is reliable and delivers PDUs in sequence Uses sequence number Application RTP RTCP Transport layer UDP IP Data Link Physical

RTP Session RTP session is sending and receiving of RTP data by a group of participants For each participant, a session is a pair of transport addresses used to communicate with the group If multiple media types are communicated by the group, the transmission of each medium constitutes a session.

RTP Synchronization Source synchronization source - each source of RTP PDUs Identified by a unique,randomly chosen 32-bit ID (the SSRC) A host generating multiple streams within a single RTP must use a different SSRC per stream

RTP Basics of Data Transmission RTP PDUs

RTP PDU Header (used by mixers) Sampling instant of first data octet multiple PDUs can have same timestamp not necessarily monotonic used to synchronize different media streams Incremented by one for each RTP PDU: PDU loss detection Restore PDU sequence Payload type Identifies synchronization source (used by mixers) Identifies contributing sources

Mixer RTP mixer - an intermediate system that receives & combines RTP PDUs of one or more RTP sessions into a new RTP PDU Stream may be transcoded, special effects may be performed. A mixer will typically have to define synchronization relationships between streams.Thus… Sources that are mixed together become contributing sources (CSRC) Mixer itself appears as a new source having a new SSRC

Translator end system 1 end system 2 transl.1 transl.2 An intermediate system that… Connects two or more networks Multicasting through a firewall Modifies stream encoding, changing the stream’s timing Transparent to participants SSRC’s remain intact end system 1 end system 2 transl.1 from ES1: SSRC=6 from ES2: SSRC=23 transl.2 authorized tunnel firewall

RTP Control Protocol (RTCP) RTCP specifies report PDUs exchanged between sources and destinations of multimedia information receiver reception report sender report source description report Reports contain statistics such as the number of RTP-PDUs sent, number of RTP-PDUs lost, inter-arrival jitter Used by application to modify sender transmission rates and for diagnostics purposes

RTCP message types Typically, several RTCP PDUs of different types are transmitted in a single UDP PDU

Sender/Receiver report PDUs RC PT=200/201  SR/RR Length (16 bits) Header SSRC of Sender RTP Timestamp Sender’s PDU Count NTP Timestamp, most significant word NTP Timestamp, least significant word Sender’s Octet Count Sender Info SSRC_1 (SSRC of the 1st Source) Fraction Lost Cumulative Number of PDU Lost Extended Highest sequence Number Received Report Block 1 Interarrival Jitter Last SR (LSR) Delay Since Last SR (DLSR) Report Block 2 SSRC_2 (SSRC of the 2nd Source) … … Profile-Specific Extensions

Ethereal capture for RTP-PDU Basic header

Ethereal capture for RTCP-PDU header of SR report sender info receiver report block SDES items

Synchronization of streams using RTCP RTP audio RTCP audio RTP video RTP video Internetwork Timestamps in RTP PDUs are tied to the individual video and audio sampling clocks timestamps are not tied to the wall-clock time, or each other! Each RTCP sender-report PDU contains (for most recently generated PDU in associated RTP stream): The timestamp of RTP PDU The wall-clock time for when PDU was created Receivers can use this association to synchronize the playout of audio and video

RTCP bandwidth scaling Problem What happens when there is one sender and many receivers? RTCP reports scale linearly with the number of participants and would match or exceed the amount of RTP data! More overhead than useful data! Example Suppose one sender, sending video at a rate of 2 Mbps. Then RTCP attempts to limit its traffic to 100 Kbps. The 75 kbps is equally shared among receivers: With R receivers, each receiver gets to send RTCP traffic at 75/R kbps. Sender gets to send RTCP traffic at 25 kbps. Solution RTCP attempts to limit its traffic to 5% of the session bandwidth to ensure it can scale! RTCP gives 75% of this rate to the receivers; and the remaining 25% to the sender.

Real-Time Streaming Protocol (RTSP) Application layer protocol (default port 554) Usually runs on RTP for stream & TCP for control Provides the control channel Uses out-of-band signaling Usable for Live broadcasts / multicast Also known as “Network remote control” for multi-media servers.

presentation descriptor RTSP Overview web browser media player Web Server Web Server/Media server HTTP presentation descriptor Presentation descriptor RTSP pres. desc,streaming commands RTP/RTCP audio/video content

RTSP Methods OPTIONS C  S determine capabilities of server/client DESCRIBE get description of media stream ANNOUNCE C  S announce new session description SETUP create media session RECORD start media recording PLAY start media delivery PAUSE pause media delivery REDIRECT redirection to another server TEARDOWN immediate teardown SET_PARAMETER change server/client parameter GET_PARAMETER read server/client parameter

RTSP Session RTSP server RTSP TCP client data source UDP media server Default port 554 media server RTSP SETUP media player RTSP server RTSP OK TCP UDP RTSP PLAY RTSP client RTSP OK RTSP TEARDOWN get UDP port RTSP OK data source choose UDP port RTP VIDEO AV subsystem RTP AUDIO RTCP

Example:Media on demand (Unicast) Media server A audio.example.com Media server V video.example.com Web server W -holds the media descriptors Client C

RTSP Message sequence W V C A C -> W : GET/Twister.sdp HTTP/1.1 Host: www.example.com Accept: application/sdp W-> C : HTTP/1.0 200 OK Content-Type: application/sdp C W V A C-> A : SETUP rtsp://audio.example.com/twister/audio.en RTSP/1.0 Cseq:1 Transport : RTP/AVP/UDP;unicast;client_port=3056-3057 A-> C : RTSP/1.0 200 OK Session: 12345678 server_port=5000-5001 C->V : SETUP rtsp://video.example.com/twister/video.en RTSP/1.0 Cseq:1 Transport : RTP/AVP/UDP;unicast;client_port=3058-3059 A-> C : RTSP/1.0 200 OK Session: 23456789 server_port=5002-5003

RTSP Message sequence (contd.) C->V: PLAY rtsp://video.example.com/twister/video RTSP/1.0 Cseq: 2 Session: 23456789 V->C: RTSP/1.0 200 OK RTP-Info: url=rtsp://video.example.com/twister/video; seq=12312232; C W V A C->A: PLAY rtsp://audio.example.com/twister/audio.en RTSP/1.0 Cseq: 2 Session: 12345678 A->C: RTSP/1.0 200 OK RTP-Info: url=rtsp://audio.example.com/twister/audio.en; seq=876655;

RTSP Message sequence (contd.) C->A: TEARDOWN rtsp://audio.example.com/twister/audio.en RTSP/1.0 Cseq: 3 Session: 12345678 A->C: RTSP/1.0 200 OK C W V A C->V: TEARDOWN rtsp://video.example.com/twister/video RTSP/1.0 Cseq: 3 Session: 23456789 V->C: RTSP/1.0 200 OK

References [1] B. A. Forouzan, “TCP/IP Protocol Suite”, Third edition, [2] H. Schulzrinne, S. Casner, R. Frederick and V. Jacobson, "RTP: a transport protocol for real-time applications", RFC 3550, July 2003. [3] H. Schulzrinne, A. Rao and R. Lanphier, "Real Time Streaming Protocol (RTSP)", RFC 2326, April 1998.

RTCP compound PDU SR SDES compound PDU (single UDP datagram) sender report receiver SSRC source 2 source 3 RTCP PDU 1 SDES CNAME PHONE RTCP PDU 2 compound PDU (single UDP datagram)

Example source 1 reports, there are 2 other sources SR sender report receiver SSRC source 2 source 3 RTCP PDU

RTCP processing in Translators SR sender information : Does not generate their own sender information(most of the times), but forwards the SR PDUs received from one side to other RR reception report blocks : Does not generate their own RR reports (most of the times), but forwards RR reports received from one side to another. SSRC are left intact SDES : Forwards without changing the SDES info. but may filter non CNAME SDES, if bandwidth is limited BYE : Forwards BYE PDU unchanged. A translator about to cease forwarding, send a BYE PDU to each connected nodes

RTCP processing in Mixers SR sender information : Generates its own SR info. Because the characteristics of source stream is lost in the mix. The SR info is sent in same direction as the mixed stream RR reception report blocks : Generates its own reports for sources in each cloud and sends them only to same cloud SDES : Forwards without changing the SDES info. but may filter non CNAME SDES, if bandwidth is limited BYE : Forwards BYE PDU unchanged. A mixer about to cease forwarding, send a BYE PDU to each connected nodes

Source description PDUs May contain: a CNAME item (canonical identifier/name) a NAME item (real user name) an EMAIL item a PHONE item a LOC item (geographic location) a TOOL item (application name) a NOTE item (transient msg, e.g. for status) a PRIV item (private extension) Value 1 2 3 4 5 6 7 8 CNAME=1 length user and domain name