Transcript Voice over IP

Voice over IP
Andreas Mettis
University of Cyprus
November 23, 2004
Overview
What is VoIP and how it works.
 Reduction of voice quality.
 Quality of Service for VoIP

VoIP
VoIP (voice over IP - that is, voice delivered using
the Internet Protocol) is a term used in IP telephony
for a set of facilities for managing the delivery of
voice information using the Internet Protocol (IP).
In general, this means sending voice information in
digital form in discrete packets rather than in the
traditional circuit-committed protocols of the public
switched telephone network (PSTN).
How VoIP Works
The OSI Model
Analog to Digital
Voice is nothing but air vibration.
 The microphone converts this vibration
into an equivalent variation of an electrical
current.
 The amplitude of this current is measured
8000 times every second.
 Each reading is coded in binary (ones and
zeros).
 Each code is made up of 8 bits.

Codec Standards
Packet by Packet Transmission
Transport Layer
The Real-time Transport (RTP) Protocol
provides end-to-end network transport
functions suitable for applications
transmitting real-time data such as audio,
video or simulation data, over multicast or
unicast network services.
 RTP does not address resource reservation
and does not guarantee quality-of-service
for real-time services.

Transport Layer
The User Datagram Protocol (UDP),
provides a simple, but unreliable message
service for transaction-oriented services.
 Each UDP header carries both a source
port identifier and destination port
identifier, allowing high-level protocols to
target specific applications and services
among hosts.

Network Layer
The Internet Protocol (IP), is the routing
layer datagram service of the TCP/IP suite.
The IP is used to route packets from host to
host.
 The IP packet header contains routing
information and control information
associated with datagram delivery.

Data Link/ Physical Layer
The Ethernet header is attached to the VoIP
frame.
 At the Physical Layer the data are sent
from the sender to the receiver.

VoIP Packet
Reduction of voice quality
Mean Option Score


In order to assess the quality of voice
communications in the presence of impairments,
it is crucial to study the individual as well as
collective effects of the impairments and produce
quantitative measures that reflect the subjective
rating that listeners would give.
MOS is valuable in that it addresses the human
perceived experience, which is the ultimate
measure of interest.
Application Layer
Standard Codec type Rate (Kbps) Frame (ms)
MOS
G.711
G.729
PCM
CS-ACELP
64
8
10
4.43
4.18
G.723.1
ACELP
5.3
30
3.83
G.723.1
MP-MLQ
6.3
30
4.00
Application Layer
Voice Activity Detection
VAD uses the fact that two communication
partners seldom speak at the same time.
 Bandwidth saving up to 50%.
 Difficult to distinguish between ambient
noise and silence in transmission.
 Voice clipping.

Packet Size
Delay



Delay incurred in encoding (Algorithmic
delay)
Packetization delay (function of the
amount of speech data included in a
packet)
Sender to receiver delay
1) Propagation delay
2) Transmission delay
3) Queuing delay
Packet losses and Delay
Echo


Echo is caused by the reflection of signals at the
four-to-two wire hybrids. This type of echo is present
when a voice call involves a combination of VoIP
segment in the Internet and a circuit segment in the
switched telephone network.
Another cause of echo is the PC-based phones that are
equipped with a microphone and loudspeakers.
Why bother about VoIP?

MONEY,MONEY,MONEY,MONEY,MO
NEY,MONEY,MONEY,MONEY!!!!!!!!!!
Quality of Service
Algorithms used
Echo Cancellation
 Loss Recovery: Forward error correction
adds redundancy information into voice
streams for aiding the loss correction.
 Error Concealment: A replacement for a
lost packet is produced which is similar to
the original lost packet. This is possible
because voice signals exhibit large
amounts of short-term self similarity.

Worst Case Design
Advantages
 QoS is guarantee.
Disadvantages
 Too expensive.
 The utilization is very small.
RSVP



The sender sends
the PATH, which
describes the
traffic that is going
to create.
The receiver sends
the RESV, that it is
used to make
reservations at
every intermediate
node.
The RESV packets
are routed using
the Reverse Path
Algorithm.
Sender 1
PATH
R
Sender 2
R
PATH
RESV
(merged)
R
RESV
R
R
RESV
Receiver B
Receiver A
RSVP
Advantages
 It is possible to assign bandwidth reliably for
eachVoIP session.
Disadvantages
 Some resources remain not used when VoIP data
has burst character.
 The load of routers becomes high and application
to a very large scale network becomes difficult.
Virtual Private Networks
Advantages
 QoS can be high.
Disadvantages
 Utilization of the network might be low.
 Might cause starvation for other VoIP
traffic.
Differentiated Services Model
Diffserv
DSCP
DSCP – Expedited Forwarding

EF – PHB ensures a minimum departure
rate, independently of any other traffic
attempting to transit across the node.

EF – PHB provides a low loss, low jitter
assured bandwidth, end to end service
through DS domains.
DSCP - Assured Forwarding
(green, yellow, red)
• Best Effort Forwarding
Admission Control



Admission control unit makes admission
decision to the new request.
Admission Criteria is a set of conditions used to
determine if an incoming call is to be accepted.
Network QoS state and flow information are
necessary for the admission control unit.
Combination of Diffserv and Call
Admission
SIP proxy observes flow information from
the router using SNMP.
 When a SIP message arrives from the SIP
terminal it decides the acceptability of this
new call based on flow information and the
SIP message log.

Diffserv Packet Marking Rule
Green: Basic data of all communication
sessions.
 Yellow: Additional data of important
sessions.
 Red: Additional data of normal sessions.

Behavior of the System


Basic data can be protected from packet loss by
dropping additional data packet of normal
communication.
In order to guarantee quality of each session, it is
necessary to make VoIP flow less than suitable
quantity on each link of the network.
Call Admission Method
Three kind of VoIP sessions which exist in a system.
 Sessions generating data traffic
 Sessions currently in the signaling stage and generating
future traffic.
 Sessions currently in the signaling stage, but which will
terminate without generating traffic in the future because
of some kind of error.
Call Admission Method



It is impossible to determine whether the session
currently in the signaling stage will generate traffic or
terminate by future error.
The log of SIP INVITE message is used and the worst
time processing of SIP signaling is recorded to log as
TTL value for each SIP INVITE message.
TTL is the worst time to process SIP signaling and is
known from statistical data.
Conclusions




VoIP is rather easy to implement but difficult to
guarantee QoS.
The combination of Diffserv and Call Admission
provide a good mechanism for QoS for VoIP.
VoIP offers a lower QoS compared with the
PSTN, and can also offers lower costs to the
organizations and people that use it.
Still need to find better solutions for providing
QoS for VoIP.
References







[1] Athina P. Markopoulou, Fouad A. Tobagi and Mansour J. Karam,
“Assessing the Quality of Voice Communications over Internet
Backbones”, pp747-760, 2003.
[2] Xiuzhong Chen, Chunfeng Wang, Dong Xuan, Zhongcheng Li,
Yinghua Min and Wei Zhao, “Survey on QoS Management of VoIP”,
Proceedings of the 2003 International Conference on Computer
networks and Mobile Computing (ICCNMC’ 03).
[3] Masaaki Noro 1, Takahiro KIKUCHI 1, Ken-ichi BABA 2,Hideki
SUNAHARA 1,3, Shinji SHIMOJO, “QoS Support for VoIP
Traffic to Prepare Emergency”, Proceedings of the 2004
International Symposium on Applications and the Internet
Workshops (SAINTW’04)
[4] http://www.protocols.com/
[5] Dr. Christos Panayiotou lecture notes.
[6] Siemens. Information and Communications networks.
[7] HiPath 4000 V1.0, IP Distributed Architecture,
Service Manual.