Transcript Chapter 29t
Instructor: Prof. Hall
Conducted by:
Wayde
Anwar
Vikas
Choong
Nathan
Nadia
29.1 Introduction
This chapter focuses on real-time
audio/video transfer over IP networks.
It examines the question of how IP can be
used to provide commercial telephone
service.
It examines the question of how routers in
an IP network can guarantee sufficient
service to provide HiQ A/V reproduction.
29.2 Audio Clips and Encoding
Standards
Simplest digitizing A/D (encoding) -> IP
network-> D/A (decoding)
– OK for audio clips, not for interactive b/c of
delay introduced
– HiQ codec are available (Amplitude overtime
to sequence of digits, reconstruct from the
digits to waveform).
– Standards: based on the tradeoffs b/w quality
and reproduction and size of digital
representation.
29.2 Audio Clips and Encoding
Standards(Continued)
– E.g. PCM for phone line (huge file production)
– Three ways to reduce the size
Produce
data at
2.2 kbps
» Fewer samples: Low quality
» Fewer bits: Low quality
» Compression: Delay(require fast CPU) good when
delay is not important.
29.3 Audio and Video
Transmission and Reproduction
AV application are real-time: timely
transmission (missing data is skipped).
How can a network guarantee that the
stream is delivered at exactly the same rate
that the sender used?
– Telephone system way: the entire system is
engineered(digital circuits included) to deliver
output at the same rate of the input even for
multiple paths.
29.3 Audio and Video Transmission and
Reproduction(cont)
– IP network is not isochronous for the delay
introduced – vary delay is called jitter.
– Additional protocol is needed in addition to IP;
» Each packet must have timestamp to tell the sender
when to play back.
» This is important b/c it tells the receiver to pause
when a packet is lost or sender stops encoding.
29.4 Jitter and Playback Delay
How
can a receiver recreate a signal
accurately if the network introduces a
jitter?
– Playback buffer (similar to queue)
– How does it work?
» The receiver introduces a delay until the buffer is filled with
incoming data (Threshold-playback point) – figure 29.1- (K) is
the unit of time of data to be played.
» The receiver plays K time units.
» If no jitter , datagrams continue to arrive at the same rate, so
the buffer is filled with K time units of un-played data
29.4 Jitter and Playback
Delay(Cont)
» If small delay, playback won’t be affected, the buffer
decreases as data are extracted, playback continues
for K units, once the delayed datagrams arrive
buffer will be refilled.
» If a datagram is lost, buffer will be empty, output
pauses for time corresponding for the missing data.
K is small – needed buffer will be used before delayed
data arrive.
K is too large – immunity to jitter with noticeable delay (in
addition to NW delay) to user.
» Playback is still used despite disadvantages.
29.5 Real-Time Transport
Protocol (RTP)
It does not provide timely transmission.
Timely manner depends on the underlying
system.
It provides:
– Sequence Number
– Timestamp
For the receiver to
control playback.
RTP does not distinguish b/w types of data;
therefore, it does not enforce uniform
interpretation of semantics.
29.5 Real-Time Transport
Protocol (RTP) (Cont)
RTP header
provides needed
information for interpretation by the
receiver:
– 2 bit version (current 2)
– 16 bit SEQUENCE NUM: first one is randomly
chosen.
– X-bit is used to identify if the application defines
optional header extension b/w RTP header and pay
load.
– 7 bit PTYPE: determines the interpretation of the most
remaining header field (Pay Load Type).
29.5 Real-Time Transport
Protocol (RTP) (Cont)
– P-bit specify whether padding is in effect to the
pay load. (Encryption: How data is allocated in
blocks).
– M-bit used by the application (Marking points –
e.g. beginning of video stream)
– 32-bit TIMESTAMP – affected by the type at
which first octet is digitized.
29.6 Streams, Mixing, and
Multicasting
Key Part to RTP is its support for
translation or mixing.
– Translation: changing the encoding of a stream
at an intermediate station.
– Mixing: receiving streams of data from
multiple sources, combining them into a single
stream, and sending the results.
» Mixers are important to service multiple streams in
conferencing.
29.6 Streams, Mixing, and
Multicasting(Cont.)
The field SYNCHRONIZATION SOURCE
IDENTIFIER specifies. Each source must
choose a unique identifier. If mixer is
enabled, the mixer will be the source of the
new stream.
The original source is not lost b/c mixer
uses CONTRIBUTING SOURCE ID to
identify the actual stream source.
CC-field gives the number of contributing
sources.
29.6 Streams, Mixing, and
Multicasting(Cont.)
RTP works with IP multicasting and mixing
especially in multicast environment.
For example, in teleconference situation,
unicast is cumbersome; however,
multicasting will allow multi-users to
communicate both ways at the same time.
Mixers make this possible by reading
several inputs resulting in fewer datagrams.
29.7 RTP Encapsulation
RTP is a transport-level protocol working
on the top of UDP.
This means that it needs to be encapsulated
in UDP before the final encapsulation in IP
datagram.
RTP does not have a reserved port number.
Port is allocated for each session, and
remote app must inform about port number.
RTP prefer even numbers.
29.8 RTP Control Protocol
So far, Real-Time transmission has been
explained as a protocol allowing
reproduction of A/V data.
Monitoring the underlying network is as
important as the protocol itself during each
session, and providing out of band com b/w
endpoints. (Adaptive applications).
An application may adjust the buffer size, or
choose lower band width due to NW cong.
29.8 RTCP (Cont.)
Out of Band can be used to send
information in parallel with real time like
caption.
RTP control protocol (RTCP) provides the
needed control functionality.
RTCP: allows senders and receivers to
transmit a series of reports one to another
that contain additional info about data
transferred in addition to NW performance.
RTCP is encapsulated in UDP using a port
number that is greater than RTP port
number.
29.9 RTCP Operation
Uses 5 basic message type:
200 - Sender Report - provides absolute
timestamp
– Absolute timestamp is essential to synchronize
multiple streams
– Since RTP require separate stream for each media ,
transmission of video/audio require 2 streams
201 - Receiver Report - Inform source about
conditions of reception
– allow participating receivers & senders in a session to
learn about reception conditions of other receivers
29.9 RTCP Operation
– allow receivers to adapt their rate of reporting to
avoid using excessive bandwidth & overwhelming
the sender
202 - Source Desc. Message - general info
about user (owns/ control source)
– Each message contain 1 section for each outgoing
RTP stream
203 - Bye Message - Shutting down a stream
204 - Application Specific Message - extend
basic facility to allow application to define
message type
29.10 IP telephony & Signaling
Real-time transmission: use of IP as the
foundation for telephone service
Researches are investigation 3 components to
replace isochronous systems:
– RTP is needed to transfer a digitized signal across
an IP internet correctly
– Mechanism is needed to establish and terminate
telephone calls
– Researches are exploring ways an IP internet can
function like an isochronous network
29.10 IP telephony & Signaling
Telephone industry use Signaling : process of
establishing a telephone call
Public Switched Telephone Network (PSTN)
uses Signaling System 7 (SS7)
– performs call routing before any audio is sent
– handles call forwarding and error conditions
29.10 IP telephony & Signaling
Signaling functionality must be available before
IP can be used to make calls
IP telephony must be also compatible with extant
telephone standards
Must be possible for IP telephony system to
interoperate with the conventional phone system
at all levels.
29.10 IP telephony & Signaling
The general approach to interoperability
uses a gateway between IP & conventional
phone system
Standards for IP Telephony:
– ITU has defined a suite or protocol known as
H.323
– IETF has proposed a signaling protocols know
as SIP
29.10.1 H.323 Standards
Originally created to allow the transmission of
voice over local area
Then it was extended to allow transmission of
voice over IP internets
Specifies how multiple protocols can be
combined to form functional IP telephony
Defines gateways & gatekeepers :
– provide a contact point for telephones using IP.
– Each IP Telephone must register with a gatekeeper
29.10.1 H.323 Standards
H.323 relies on 4 major protocols:
–
–
–
–
H.225.0 Signaling used to establish a call
H.224 Control and feedback during the call
RTP Real-time data transfer
T.120 Exchange of data associated with a call
Fig 29.5 illustrates relationship among the
H.323 protocols
29.10.2 Session Initiation
Protocol (SIP)
Covers only signaling, doesn't supply all of
H.323 functionality
Uses client-server interaction, with servers
being divided into 2 types:
– user agent server runs in a SIP telephone
» assigned an identifier: user@site
– intermediate server ; between 2 SIP telephone
» handles call set up and call forwarding
29.10.2 Session Initiation
Protocol (SIP)
SIP relies on Session Description Protocols
SDP (companion protocol)
SDP important in conference call
– participants join and leave dynamically
SDP specifies media encoding, protocols
number and multicast address
29.11 Resource
Reservation/Quality of Service
Quality of Service (QoS) refers to statistical
performance guarantees
– regarding loss, delay, jitter and throughput
An isochronous network that meet strict
perfomacnce bounds provide QoS
Packet switched network doesn't provide
QoS
Is QoS needed for real-time transfer of
voice & video over IP?
29.11 Resource
Reservation/Quality of Service
Internet send audio but operates without
QoS
ATM, derived from telephone system
model, provide QoS guarantees
IETF adopted a differentiated services
approach
– divide traffic into separate QoS classes
– sacrifice fine grain control for less complex
forwarding
29.12 QoS Utilization &
Capacity
Central issue is utilization
– a network with 1% utilization: doesn’t need
QoS
– a network with 1o1% utilization: will fail under
any QoS
Proponent who argue for QoS assert that
QoS mechanism is important because:
– by dividing the existing resources among more
users, system become more “fair”
29.12 QoS Utilization &
Capacity
– by shaping traffic, the network run at higher
utilization without danger of collapse
As long as rapid increases in capacity
continues, QoS represent cause unnecessary
overhead
When demand rises more rapidly than
capacity, it becomes an economic issue
29.13 RSVP
How can IP network provide QoS?
IETF produced 2 protocols: RSVP & COPS
QoS cannot be added at the application layer to
IP; basic infrastructure must change
Infrastructure must change: routers must agree
to reserve resources
Endpoints must send a request to spefiicy
resources needed before data is sent
As datagrams traverse the flow, routers need to
monitor (traffic policing) and control traffic
forwarding
29.13 RSVP
Control of queuing is needed:
– router must implement a queuing policy that meets
guaranteed bounds on delay
– router must smooth packet burst (traffic shaping)
RSVP is not a routing protocols; operates
before any data is sent and handles reservations
request and replies.
RSVP is unidirectional (simplex); if application
needs QoS in two directions, each point must
use RSVP to request a separate flow
29.14 COPS
When an RSVP arrivers a router must evaluate:
– feasibility : a local decision
– policies: requires global cooperation
IETF architecture uses 2-level model:
» when router receiver RSVP request, it becomes a client
which consult server :Policy Decision Point (PDP) to
determine whether request meets policy constraints
» if PDP approves a request, router must operate as Policy
Point Point (PEP)to ensure traffic does not exceed the
approved policy
COPS protocol define the client-server
interaction between a router and a PDP
29.14 COPS
Although COPS defines it own message
header, the underlying format shares many
details with RSVP
When a router receives an RSVP request:
– extract items related to policy
– place them in a COPS message
– send the result to PDP
Summary
Audio data can be encoded in digital form
(hardware:codec)
Pulse Code Modulation (PCM) produce
digital values at 64 Kbps
RTP is used to transfer real-time data across
an IP internet. Each message contain :
– sequence number
– a media timestamp
Summary
RTCP is used to supply information about sources
& allow mixer to combine several streams
Debate continues where Q0S guarantees is
needed to provide real-time
Endpoints use RSVP to request a flow with
specific QoS; intermediate routers either approve
or deny the request
When RSVP request arrives, router use COPS to
contact PDP and verify that request meets policy
constraints