1. RTP: Real-time Transport Protocol
By Igor Medvetsky

2. OSI Model and protocol environment

3. TCP vs. UDP TCP features:
applications on networked hosts create a connection one to another
guarantees reliable and in-order delivery of sender to receiver data
sequence numbers for ordering received TCP segments and detecting duplicate data
checksums for segment error detection
acknowledgements and timers for detecting and adjusting to loss or delay
retransmission and timeout mechanisms for error control
unpredictable delay characteristics
Hence: not suitable for real-time communication

4. TCP vs. UDP UDP features:
simple, unreliable datagram transport service
does not provide reliability and ordering guarantees
datagrams may arrive out of order or go missing without notice
checksum for detecting packages containing bit errors
faster and more efficient for many lightweight or time- sensitive purposes
obvious choice for real-time video transmission

5. Overview of RTP
RTP is the Internet-standard protocol for the transport of real-time data, including audio and video.
media-on-demand
Internet telephony.
RTP consists of a data and a control part. The latter is called RTCP (RTP Control Protocol).
The data part of RTP is a thin protocol providing support for applications with real-time properties such as continuous media, including:
timing reconstruction
loss detection
security and content identification.

6. Overview of RTP
RTCP provides support for real-time conferencing of groups of any size within an internet.
It offers quality-of-service feedback from receivers
support for the synchronization of different media streams.
Reference: RFC 3550, July 2003

7. General Scenario One-to-one One-to-many Many-to-many
Local transmission (access within one machine)
RTP packets
RTCP (Sender and Receiver Reports)

8. RTP packets
Consist of and RTP header, optional payload headers and the payload itself
RTP overhead = 12 bytes
IP+UDP+RTP overhead = = 40 bytes
It is advisable to keep coded slice sizes as close to, but never bigger than, the MTU(Maximum transmission unit) size
MTU sizes: ~1500 bytes for wireline IP links (max. size of an Ethernet packet), ~100 bytes in wireless environments

9. RTP packet header

10. Real-Time Control Protocol

11. Real-Time Control Protocol
Periodic transmission of control packets to all participants in the session, using the same distribution mechanism as the data packets.
The underlying protocol must provide multiplexing of the data and control packets, for example using separate port numbers with UDP.
It is recommended that the fraction of the session bandwidth allocated to the RTCP is 5%.

12. Real-Time Control Protocol

13. RTCP packets
SR – sender report, for transmission and reception statistics from participants that are active senders
RR - Receiver report, for reception statistics from participants that are not active senders
SDES - Source description items
BYE - Indicates end of participation
APP - Application specific functions

14. Feedback in RTCP
Using the SR and RR information we can obtain the following measurements:
Packet loss rate over the interval between two reception reports
Number of packets expected during the interval
Packet loss fraction over the interval
Loss rate per second
Number of packets received
Statistical validity of any loss estimates
interarrival jitter
Timestamps allowing to calculate the Round-Trip Time delay

15. The biasing algorithm

16. The biasing algorithm
The biasing algorithm provides a way of balancing workload by specifying that workload distribution should favor particular regions. For example, if there are two regions with a bias of 75 and 25, program requests are sent in a ratio of 3:1 to the first region.

17. Analysis of bandwidth redistribution algorithm for single source multicast
Dan Komosny, Vit Novotny

18. Goals of the article
Analysis made for the one-to-many multicast sessions, and is used to increase the feedback rate of selected session members above a minimum guaranteed value.
A higher feedback rate allows reporting more accurate results.
Using the spare bandwidth to increase the feedback rate for the VIP group members.

19. A little reminder about RTCP (if you forgot from previous slides)
RTCP provides a set of messages exchanged among session members
These messages are used as feedback for monitoring the session behavior
The session feedback could be used for:
the parameterization of a multicast forward error correction (FEC) algorithm
tuning of algorithms
summary of data sent
synchronization of different transmitted media
packet loss
packet delays
interarrival jitter.
localize distribution problems for particular receivers.

20. However,
Not all the feedback data from session members are of equal importance

21. The two known methods for RTCP
Reflection method
Summarization method
the source forwards every RR packet to all session receivers.
could be harmful to the network load.
does not need to forward all the received data
aggregating selected received data from RR packets in the source.
summary packet is assembled and sent to all receivers via multicast
aggregated values can be compressed up to a factor of 16.

22. What is the problem?
The session control is thus strongly affected by the frequency of feedback reporting
The RTCP packets (either SR or RR) should therefore be sent as often as possible
On the other hand, many RTCP packets can load the assigned session channel too much
Practically, the minimum transmission interval used between RR packets is 0.2 pkt/sec

23. Authors divide receivers and their feedback into 2 groups
important (VIP)
unimportant (remainders - REM)

24. Authors divide receivers and their feedback into 2 groups
Member feedback from the VIP group is more important to receive than the feedback from the other members
VIP members must have minimum packet rate (0.01 pkt/sec)
VIP group has a constant number of members whereas the number of remaining members varies
For setting relation between VIP and remainders authors use biasing algorithm. VIP and the remaining members has a weight ratio of 1:10

25. ניתוח: נוסחה כללית לחישוב קצב השידור :
PR = packet rate , PS = packet size
כעת נגדיר 3 ערכים עבור כמות משתמשים בקבוצת remainders : nrem0, nrem1 and nrem2
nrem0 –כאשר קצב חבילות של הקבוצה מתחיל לרדת מתחת ל- 0.2 pkt/sec
nrem1 - כאשר קצב חבילות של הקבוצת VIP מתחיל לרדת
nrem2 - כאשר קצב חבילות של הקבוצת VIP מגיע ל- PRvip_min.
( * PRcrit is the feedback rate of the remaining session members when the feedback rate PRvip of VIP group starts to decrease from PRmax(
כמו כן חשוב להדגיש כי מספר איברי VIP בקבוצה צריך לקיים:

26. ניתוח:
עכשיו ניתן לחשב את קצב שידור החבילות ברשת:
PRrr

27. דוגמה ותוצאות
ישנם שני מקרים עבור nrem2 :

28. דוגמה ותוצאות

29. מסקנות
לפעמים עדיף לבחור קבוצה מייצגת או קבוע של multicast receivers ולקבל מהם feedback איכותי ומלא מאשר לקבל feedback חלקי מכל המשתמשים.

30. שאלות ?