Content uploaded by Shaimaa Abbas Sharafali
Author content
All content in this area was uploaded by Shaimaa Abbas Sharafali on Oct 21, 2023
Content may be subject to copyright.
Comparative Study of MPLS Traffic Engineering
Signal Protocols with Different Audio Codecs
A Thesis
Submitted to the College of Engineering
of Al-Nahrain University in a Partial Fulfillment
of the Requirements for the Degree of
Master of Science
in
Computer Engineering
by
SHAIMAA ABBAS SHARAFALI
(B.Sc. 2011)
Ramadan 1436
July 2015
I
Abstract
Multiprotocol Label Switching (MPLS) technology is capable of reducing network
congestion and enhancing management of network resources by adapting efficient
load balancing schemes based on Traffic Engineering (TE) signal protocols. This
thesis addresses the performance of MPLS networks in the presence of TE signal
protocols for Voice over Internet Protocol (VoIP) applications.
Performance comparison between two TE signal protocols, namely the
Constraint-based Routing Label Distribution Protocol (CR-LDP) and Resource
Reservation Protocol (RSVP), is made. Simulation results are presented of two four-
switch MPLS networks having six and eight routers, respectively. The study
addresses the following different audio codecs including PCM (64 Kbps), GSM FR
(13 Kbps), G.723.1 (5.3 Kbps), G.726 (16 Kbps), G.728 (16 Kbps), G.729 (8 Kbps)
and IS-641(7.4 Kbps). The performance of these codecs is presented and compared
in the absence and presence of Quality of Service (QoS) algorithms including First-
In First-Out (FIFO), Priority Queuing (PQ), and Weighted Fair Queuing (WFQ).
Numbers of maintained calls, jitter, Packet Delay Variation (PDV), and end-to-end
delay are used as measure parameters for performance assessment.
Before using QoS algorithms, the MPLS network with CR-LDP has a
noticeable performance advantage compared to the MPLS network with RSVP. It is
(500%) more than RSVP in terms of number of maintained calls in the first network
and (700%) in the second network. After applying QoS algorithms, the WFQ queuing
is more compatible with RSVP than CRLDP protocol. The number of maintained
calls in RSVP-WFQ is (228%) more than in CRLDP-WFQ for the first network and
(107%) for the second network. Using QoS along with TE in MPLS network
decreases the jitter, packet delay variation, and end-to-end packet delay compared to
the case of using TE alone. Further, the PQ algorithm gives the lowest delay of the
other algorithms. Comparison among audio codecs shows that G.723.1 gives the
highest number of calls when RSVP, WFQ, and G.723.1 are used together.
Simulation results are obtained using OPNET modeler 14.5.
II
List of Contents
Page
Contents
I
Abstract
II
List of Contents
V
List of Symbols
VI
List of Abbreviations
IX
List of Tables
X
List of Figures
Chapter One
1
Introduction
1
Overview
3
1.2 Motivation
5
1.3 Literature Survey
7
1.4 Aim of the Thesis
8
1.5 Thesis Organization
Chapter Two
9
Concepts of MPLS and QoS
9
2.1 Introduction
9
2.2 MPLS Overview
10
2.3 MPLS Technology
10
2.3.1 MPLS Header
11
2.3.2 MPLS Architecture
14
2.4 MPLS Functionality
16
2.5 MPLS Traffic Engineering
18
2.6 Signaling Protocols
III
20
2.6.1 Constraint-based Routing Label Distribution Protocol
22
2.6.2 Resource Reservation Protocol (RSVP)
24
2.7 Quality of Service
24
2.7.1 Quality of Service Models
27
2.7.2 Classification and Conditioning of Traffic in DiffServ
27
2.8 Queue Scheduling Mechanisms
31
2.9 Performance Metrics
Chapter Three
Performance Evaluation of MPLS TE Signal Protocols for Voice
Applications
3.1 Introduction
3.2 Network Setup
36
3.3 Simulation Parameters
38
3.4 Simulation Results
46
3.5 Result Summary
Chapter Four
49
Effect of QoS Algorithms on the Performance of MPLS-TE
Networks
49
4.1 Introduction
49
4.2 Results Representation
49
4.2.1 First-In First-Out (FIFO) Algorithm
4.2.2 Priority Queuing (PQ)
57
4.2.3 Weighted Fair Queuing (WFQ) Algorithm
4.3 Results
4.4 Summary of Results and Discussion
IV
65
4.4.1 Number of Maintained Calls
68
4.4.2 Voice Jitter
4.4.3 Packet Delay Variation (PDV)
4.4.4 End-to-end Delay
4.4.5 Main Opinion Score
Chapter Five
75
Conclusions and Suggestions for Future Work
75
5.1 Conclusions
76
5.2 Suggestions for Future Work
78
References
APPENDIX A
A-1
Voice over Internet Protocol (VoIP)
A-1
A.1 VoIP
A-1
A.2 VoIP codecs
A-3
A.2.1 G.711 (PCM)
A-3
A.2.2 G.723
A-4
A.2.3 G.726
A-4
A.2.4 G.728
A-4
A.2.5 G.729
A-5
A.2.6 GSM FR
A-5
A.2.7 IS-641
V
List of Symbols
Symbols
Notations
End-to-end delay
Network delay
Encoding delay
Decoding delay
Compression delay
Decompression delay
Effect of impairments that occurs with
the voice signal
Impairments caused by different types
of losses occurring due to codec's and
network
Impairment caused by delay
particularly mouth-to-ear delay
VI
List of Abbreviations
Abbreviation
Description
3GPP
3rd Generation Partnership Project
AF
Assured Forwarding
ATM
Asynchronous Transfer Mode
BA
Behavior Aggregate
BE
Best Effort
BGP
Border Gateway Protocol
CBR
Constraint-Based Routing
CR-LDP
Constraint-Based Routing Label Distribution Protocol
CR-LSP
Constraint Label Switched Path
CSPF
Constrained Shortest Path First
CoS
Class of Service
DiffServ
Differentiated Service
DSCP
Differentiated Service Code Point
DSP
Digital Signal Processing
EF
Expedited Forwarding
E-mail
Electronic-Mail
ER-LSP
Explicit Route- Label Switched Protocol
Exp
Experimental
FEC
Forwarding Equivalence Class
FIB
Forwarding Information Base
FIFO
First-In First-Out
FTP
File Transfer Protocol
GMPLS
Generalized Multiprotocol Label Switching
GSM-FR
Global System for Mobile Communications-Full Rate
HTTP
Hypertext Transfer Protocol
VII
IETF
Internet Engineering Task Force
IP
Internet Protocol
IPv4
Internet Protocol version 4
IPv6
Internet Protocol version 6
IntServ
Integrated Service
IS-IS
Intermediate System-to-Intermediate System
LDP
Label Distribution Protocol
LER
Label Edge Router
LFIB
Label Forwarding Information Base
LIB
Label Information Base
LSP
Label Switched Path
LSR
Label Switched Router
Mbps
Mega bit per second
MPEG
Moving Picture Expert Group
MOS
Main Opinion Score
NS
Network Simulator
OPNET
Optimum Network Engineering Tools
OSPF
Open Shortest Path First
PCM
Pulse Code Modulation
PDV
Packet Delay Variation
PHB
Per-Hop Behavior
PQ
Priority Queuing
QoS
Quality of Service
QualNet
Qual Network
RESV
Reservation
RFC
Request for Comment
RIB
Routing Information Base
VIII
RSB
Reservation State Block
RSVP
Resource Reservation Protocol
RTCP
Real Time Control Protocol
RTP
Real Time Protocol
S
Stack
SLA
Service Level Agreement
TCA
Traffic Conditioning Agreement
TCP
Transmission Control Protocol
TE
Traffic Engineering
ToS
Type of Service
TTL
Time-to-Live
VC
Virtual Circuit
VoIP
Voice over Internet Protocol
VPN
Virtual Private Network
WFQ
Weighted Fair Queuing
IX
List of Tables
Table
Legend
Page
1-1
Comparison between IP, ATM, and MPLS technologies
4
2-1
A basic comparison between Hop-by-Hop and Explicit routing LSPs
14
3-1
Application parameters
3
3-2
Performance Metrics for CR-LDP and RSVP Signal Protocols based networks
45
4-1
Traffic drop time (sec)
62
4-2
Number of calls/sec
62
4-3
Voice Jitter
63
4-4
Packet Delay Variation
63
4-5
End to End Delay
64
4-6
Main Opinion Score (MOS) values
64
5-1
The proportion of Number of calls
76
A-1
Delay specifications
A-2
X
List of Figures
Figure
Legend
Page
1-1
Connectionless and connection oriented networks
2
2-1
MPLS forwarding label between routers
10
2-2
MPLS header building block
11
2-3
LSR architecture
12
2-4
LER architecture
12
2-5
Flow of packet in FEC
13
2-6
Control and Forwarding components in MPLS
15
2-7
MPLS Label Operations
16
2-8
CR-LDP operation
20
2-9
CR-LDP initiation
21
2-10
RSVP operation
23
2-11
DiffServ aggregation behavior
26
2-12
DiffServ Architecture
26
2-13
FIFO Queue concept
28
2-14
Priority queuing concept
29
2-15
Weighted fair queuing concept
30
3-1
First MPLS network topology
35
3-2
Second MPLS network topology
35
3-3
Example of results presentation
40
3-4
Scenario PCM codec in the 1st network
41
3-5
Scenario PCM codec in the 2nd network
41
3-6
Scenario GSM FR codec in the 1st network
41
3-7
Scenario GSM FR codec in the 2nd network
41
3-8
Scenario G.723.1 codec in the 1st network
42
3-9
Scenario G.723.1 codec in the 2nd network
42
3-10
Scenario G.726 codec in the 1st network
42
3-11
Scenario G.726 codec in the 2nd network
42
XI
3-12
Scenario G.728 codec in the 1st network
43
3-13
Scenario G.728 codec in the 2nd network
43
3-14
Scenario G.729 codec in the 1st network
43
3-15
Scenario G.729 codec in the 2nd network
43
3-16
Scenario IS-641 codec in the 1st network
44
3-17
Scenario IS-641 codec in the 2nd network
44
3-18
Number of maintained calls/sec in both networks
46
3-19
Voice jitter for both networks
47
3-20
Voice PDV for both networks
47
3-21
Voice end-to-end delay for both networks
48
3-22
MOS performance in both networks
48
4-1
Scenario PCM codec with applying FIFO algorithm in the 1st network
50
4-2
Scenario PCM codec with applying FIFO algorithm in the 2nd network
50
4-3
Scenario GSM FR codec with applying FIFO algorithm in the 1st network
50
4-4
Scenario GSM FR codec with applying FIFO algorithm in the 2nd network
50
4-5
Scenario G.723.1 codec with applying FIFO algorithm in the 1st network
51
4-6
Scenario G.723.1 codec with applying FIFO algorithm in the 2nd network
51
4-7
Scenario G.726 codec with applying FIFO algorithm in the 1st network
51
4-8
Scenario G.726 codec with applying FIFO algorithm in the 2nd network
51
4-9
Scenario G.728 codec with applying FIFO algorithm in the 1st network
52
4-10
Scenario G.728 codec with applying FIFO algorithm in the 2nd network
52
4-11
Scenario G.729 codec with applying FIFO algorithm in the 1st network
52
4-12
Scenario G.729 codec with applying FIFO algorithm in the 2nd network
52
4-13
Scenario IS-641 codec with applying FIFO algorithm in the 1st network
53
4-14
Scenario IS-641 codec with applying FIFO algorithm in the 2nd network
53
4-15
Scenario PCM codec with applying PQ algorithm in the 1st network
54
4-16
Scenario PCM codec with applying PQ algorithm in the 2nd network
54
4-17
Scenario GSM FR codec with applying PQ algorithm in the 1st network
54
4-18
Scenario GSM FR codec with applying PQ algorithm in the 2nd network
54
4-19
Scenario G.723.1 codec with applying PQ algorithm in the 1st network
55
4-20
Scenario G.723.1 codec with applying PQ algorithm in the 2nd network
55
XII
4-21
Scenario G.726 codec with applying PQ algorithm in the 1st network
55
4-22
ScenarioG.726 codec with applying PQ algorithm in the 2nd network
55
4-23
Scenario G.728 codec with applying PQ algorithm in the 1st network
56
4-24
Scenario G.728 codec with applying PQ algorithm in the 2nd network
56
4-25
Scenario G.729 codec with applying PQ algorithm in the 1st network
56
4-26
Scenario G.729 codec with applying PQ algorithm in the 2nd network
56
4-27
Scenario IS-641 codec with applying PQ algorithm in the 1st network
57
4-28
Scenario IS-641 codec with applying PQ algorithm in the 2nd network
57
4-29
Scenario PCM codec with applying WFQ QoS algorithm in the 1st network
58
4-30
Scenario PCM codec with applying WFQ algorithm in the 2nd network
58
4-31
Scenario GSM FR codec with applying WFQ algorithm in the 1st network
58
4-32
Scenario GSM FR codec with applying WFQ algorithm in the 2nd network
58
4-33
Scenario G.723.1 codec with applying WFQ algorithm in the 1st network
59
4-34
Scenario G.723.1 codec with applying WFQ algorithm in the 2nd network
59
4-35
Scenario G.726 codec with applying WFQ algorithm in the 1st network
59
4-36
Scenario G.726 codec with applying WFQ algorithm in the 2nd network
59
4-37
Scenario G.728 codec with applying WFQ algorithm in the 1st network
60
4-38
Scenario G.728 codec with applying WFQ algorithm in the 2nd network
60
4-39
Scenario G.729 codec with applying WFQ algorithm in the 1st network
60
4-40
Scenario G.729 codec with applying WFQ algorithm in the 2nd network
60
4-41
Scenario IS-641 codec with applying WFQ algorithm in the 1st network
61
4-42
Scenario IS-641 codec with applying WFQ algorithm in the 2nd network
61
4-43
Number of maintained calls/sec in the first network
66
4-44
Number of maintained calls/sec in the second network
66
4-45
Average number of maintained calls/sec for different QoS algorithms for both
networks
67
4-46
Average number of maintained calls/sec as codecs for various codecs
67
4-47
Jitter associated with the first network
68
4-48
Jitter associated with the second network
69
4-49
Average jitter values (msec)
69
4-50
PDV associated with the first network
70
XIII
4-51
PDV associated with the second network
71
4-52
Average PDV values (msec) of both networks
71
4-53
End-to-end delay values of the first network
72
4-54
End-to-end delay values of the second network
72
4-55
Average end-to-end delay values for different codecs
73
4-56
Average end-to-end delay values for different QoS
73
4-57
MOS performance of the first network
74
4-58
MOS performance of the second network
74
Chapter One Introduction
Chapter One
Introduction
1.1 Overview
The circuit switched networks were the early computer networks where the
physical links were used for carrying the continuous bit stream. This was
well appropriate for voice and data unicast communications but, in case of
failure, it leads to some dire consequences, all the communications over
the failed link are interrupted in such a case [1]. The packet-switched
network repairs this defect by splitting data into small parts called packets,
these packets are separately routed across a network so different packets
can take different paths. Thus if a link drops, packets can be rerouted
through alternate available links to avoid failed link and hence
communication is not interrupted [2]. This trait makes packets-switched
networks more reliably but on the other hand, as packets are routed
separately, it is complicated for management flow of data [1].
Internet Protocol (IP) network uses datagram packet switching
technique [3]. It is unacceptable to real-time applications like voice and
video because the IP router's routing table can be complicated and time
consuming. Each IP packet consists of a header and a payload, and the
header contains the IP destination address, it is a connectionless protocol
that is principally responsible for addressing and routing packets between
network devices. Connectionless means that before exchanging the data
there is no session established connection (i.e., it does not guarantee packet
delivery). This problem is solved by using Asynchronous Transfer Mode
(ATM) [4], as shown in Figure (1-1).
Chapter One Introduction
Figure (1-1): Connectionless and connection oriented networks [4].
IP networks are frequently layered over ATM networks, and that is very
costly with regard to the overhead (adding 25% or more of overhead to
every IP packet) [1]. ATM transports fixed-length (53 byte) cells and not
variable length. Five bytes are used for the header and the remaining 48 for
the payload. Packets over an ATM network should be divided, transported,
and re-collected employing an adaptation layer which adds so many
complications and overhead to the data stream [5]. The drawback of ATM
is the matter of scalability and that one link failure can lead to dozens of
Virtual Circuits (VCs) going down [6].
Internet traffic has been increased enormously, so that the amount of
traffic in 2014 reached 63.9 Exabyte (Exabyte = 260 byte) on a monthly
basis approximately, quadrupling the corresponding amount of 2009. The
traffic of real-time applications such as VoIP, video streaming, and video
conference has been dominated as it provided more than 91% of the global
consumed IP traffic. Additionally, traffic generated for business, file
sharing, and mobile data has strengthened this growth [7]. Thus real-time
applications cause a massive congestion to the IP networks, in addition to
Chapter One Introduction
bandwidth requirements it also requires other Quality of Service (QoS)
assurances, such as jitter, end-to-end delay, or packet loss probability.
These QoS requirements add new challenges on Internet service providers
[8]. The standard organizations like Internet Engineering Task Force
(IETF) has suggested several standards to achieve the QoS in the IP
networks. These include the Multiprotocol Label Switching (MPLS)
Network, and also Differentiated Services (DiffServ) where the several
Request for Comments (RFCs) have been published for these two services.
1.2 Motivation
MPLS combines the flexibility of the IP routing protocols with layer 2
switching functionality to introduce fast packet switching in frame based
IP networks [5]. The other remarkable trait of MPLS is its ability to provide
Traffic Engineering (TE) that plays an important role in decreasing the
congestion by using effective load balancing and management of the
network resources [9], and setting up label switched-paths (LSPs) along
links to available resources, thus ensuring that bandwidth is always
available for a particular flow and avoiding congestion. MPLS TE operates
at an aggregate level across all classes of service and as a result, it cannot
gives bandwidth guarantees on a per class basis [10]. However, MPLS TE
is not aware of QoS [11]. The DiffServ architecture is one of the QoS
technique types designed to provide differing levels of QoS to different
traffic flows. MPLS DiffServ-TE makes MPLS-TE aware of QoS [12]. The
benefits of combining the functionalities of TE and DiffServ are that MPLS
DiffServ-TE can deliver the QoS guarantees to meet strict Service Level
Agreements (SLAs). It performs DiffServ classification in an MPLS-TE
network, coupling the advantages of both DiffServ and MPLS-TE and
achieving all QoS functionalities. DiffServ with TE is the most advanced
technology which achieves QoS in a scalable, flexible and dynamic way,
Chapter One Introduction
satisfying all next generation network requirements [13]. Thus MPLS
overcomes the constraints such as high packet loss and excessive delays of
IP networks. Since MPLS provides scalability and congestion control,
MPLS is considered ideal for real-time applications [14]. A comparison of
IP, ATM, and MPLS is shown in Table (1-1).
Table (1-1): Comparison among IP, ATM, and MPLS technologies [15].
Parameter
IP
ATM
MPLS
Connection type
Connectionless
Connection-
oriented
Connection-oriented
Packet flow
direction
Bidirectional
Bidirectional
Unidirectional
No. of bytes
transferred
100 to 1500
bytes
Fixed 53 bytes
Variable
Switching method
Packet
switching
Circuit
switching
Both
Restoration
facility
Less
More
Highest
Efficiency
Less
Higher than IP
Highest
Delay offer
Makes delay
Lower than IP
Lowest
QoS
More than
ATM
Less than IP
Highest
Compatibility
with future
technology
Compatible
with MPLS but
not with ATM
Compatible
with MPLS but
not with IP
Compatible with IP
and ATM, compatible
with future
technology
Chapter One Introduction
1.3 Literature Survey
In 2008, Porwal et al. [4] made a comparative analysis between
MPLS and IP networks. Qual Network (QualNet) 4.0 simulator was used
to simulate the performance. The parameters considered were packet loss,
throughput and end-to-end delay. They concluded that the performance of
MPLS network outperforms the performance of IP network.
In 2008, Hodzic and Zoric [16] studied the effect of TE in MPLS
network and compared the results with that of the MPLS network without
TE. Their study used two signaling protocols CR-LDP and RSVP in MPLS
networks and they used Network Simulator-2 (NS-2) for simulating their
results. The results reveal the use of TE enhances the performance of the
network by efficient distribution of traffic load to the available links.
In 2010, Jannu and Deekonda [14] compared the MPLS-TE
network performance with that of IP network using the same network
topology. The applications used were File Transfer Protocol (FTP), video
conferencing, and voice with using Pulse Codes Modulation (PCM) audio
codec. The study compares the number of calls that the IP network can
handle with the MPLS-TE network on the basis of jitter, end-to-end delay,
and packet sent and packet received. OPNET 14.5 simulator was used to
simulate the networks. The main conclusions drawn from this study are
MPLS-TE network can handle a larger number of calls than the IP network.
Further, the number of maintained calls in MPLS-TE network was 100
calls/sec while the traditional IP network handled 70 calls/sec.
In 2011 Kharel and Adhikari [9] implemented QoS on the MPLS-
TE network. Basic scheduling algorithms of First-In First-Out (FIFO),
Priority Queuing (PQ), and Weighted Fair Queuing (WFQ) were used.
Chapter One Introduction
Performance evaluation parameters were jitter, and packet end-to-end
delay, and Packet Delay Variation (PDV). They did not consider the
number of maintained calls in their parametric study. The OPNET modeler
16.0 was used for the simulating the results. The applications of FTP, video
conferencing, and voice with PCM codec were used. They concluded that
using TE along with QoS in MPLS network decreases the jitter, PDV and
end-to-end packet delay compared to using TE alone for voice traffic.
Among these three scheduling algorithms, PQ gives the lowest jitter, PDV,
and end-to-end delay.
In 2012, Aziz et al. [17] compared the MPLS networks with Internet
Protocol version 4 (IPv4) and Internet Protocol version 6 (IPv6) Networks.
The performance parameter used in the study was PDV. The effectiveness
of the DiffServ and MPLS TE integration in IPv4/IPv6 network was
clarified and analyzed. The results show that IPv6 experiences more PDV
than that of using IPv4.
In 2012, Naoum and Maswady [18] compared the performance of
VoIP in IP network and in MPLS-TE network by using CR-LDP signal
protocol. The applications used in this study were FTP, video streaming,
and voice with G.723.1 codec type. The results indicate that MPLS TE
network has a better overall performance for voice in all voice metric used
in this study, which are number of maintained calls, end-to-end delay,
voice jitter, and PDV. The number of calls obtained from MPLS network
is three times more than that obtained from the IP network.
In 2013, Ibrahim and AL-Quzwini [19] studied the performance
of MPLS-TE network signal protocols CR-LDP and RSVP. The
applications used were Electronic mail (Email), Hypertext Transfer
Chapter One Introduction
Protocol (HTTP), FTP, video conference, print, telnet, data base, and voice
with different audio codec types including PCM, G.723.1, and G.729. The
simulation results show that the MPLS TE network with CR-LDP
outperforms the MPLS network with an RSVP in terms of the number of
calls/sec handled and the amount of received voice packets with all audio
codecs used.
In 2014, Chaudhary and Singh [20] compared the performance of
IP and MPLS networks, for voice applications. The CR-LDP protocol was
used to implement TE in MPLS network. The comparison parameters were
voice sent and received packets, throughput, voice packet loss, and packet
delay. NS-2 software was used to simulate both networks. Their results
show that IP network drops more packets and causes more delay than the
MPLS network.
This thesis compared the performance of the two TE signal protocols
in MPLS networks with applying QoS.
1.4 Aim of the Thesis
The aim of this thesis is to assess the performance of two TE signal
protocols, CR-LDP and RSVP, for voice traffic in MPLS networks. The
investigation should address the effect of applying different QoS
algorithms on network performance when different types of audio codecs
are used. The performance parameters should include number of
maintained calls/sec, jitter, PDV, end-to-end delay, and Main Opinion
Score (MOS) in order to get a clear comparison between the two TE
protocols.
Chapter One Introduction
To achieve these goals, the following steps are followed. Two
different MPLS-TE networks are used with three QoS algorithms and
seven types of audio codecs. The results are evaluated by considering the
same performance parameters as follows
i. Calculating the number of maintained calls for each case of the
two TE signal protocols under different QoS algorithms and
different audio codecs.
ii. Analyzing the performance parameters such as voice jitter, voice
PDV, MOS, and end-to-end delay.
iii. Determining which TE signal protocol performs better, which
QoS algorithm improves the performance, and which audio
codec is more compatible with MPLS network.
1.5 Thesis Organization
This thesis is organized in five chapters; the remaining chapters can be
summarized as follows
Chapter 2 provides an overview of MPLS network covering important
details such as label, traffic engineering, and signaling protocols. This
chapter also briefly introduces the main concepts of QoS.
Chapter 3 studies the performance of MPLS networks operating with TE
signal protocols and different audio codecs for voice applications. The
performance parameters cover a number of maintained calls/sec, voice
jitter, voice PDV, voice end-to-end delay, and MOS.
Chapter 4 studies the effect of applying QoS algorithms to the
performance of TE signal protocols.
Chapter 5 presents the main conclusions drawn from this study along with
suggestions for future work.
Chapter Two Concepts of MPLS and QoS
Chapter Two
Concepts of MPLS and QoS
2.1 Introduction
This chapter contains two parts to present the main concepts behind
Multiprotocol Label Switching (MPLS) and Quality of Service (QoS). The
MPLS architecture is explained in part one of this chapter to identify the
main used technologies. The MPLS functionality is also stated in this part
with emphasis being placed on the separation of control and forwarding
planes, and label swapping forwarding mechanism. Brief description of
Traffic Engineering and the two types of signaling protocols, namely
Constraint-based Routing Label Distribution Protocol (CR-LDP), and
Resource Reservation Protocol (RSVP), is also given. The second part of
this chapter describes QoS models and queue scheduling mechanisms used
in this thesis.
2.2 MPLS Overview
MPLS is a sophisticated technology for routing the packets in a data
network and provides high performance packet control and forwarding
mechanism. This technology is an expansion of IP architecture by adding
new features to the IP network, therefore it cannot be considered as a
replacement for the IP [4]. MPLS uses 20-bits for label which is added to
the packet to act as identifier for packet forwarding. Before forwarding
packets, all of them are labeled. Analysis of the packets network layer
header is not required at the downstream routers since all packets are
labeled before forwarding [21]. Label is placed in MPLS header between
Data link layer (layer 2) and Network layer (layer 3) of the IP packet to
form layer 2.5 label switched network [4]. The MPLS forwarding label is
Chapter Two Concepts of MPLS and QoS
shown in Figure (2-1). Although MPLS is a relatively simple technology,
it enables sophisticated capabilities far superior to the Traffic Engineering
function in ordinary IP networks. When MPLS is combined with
Differentiated Services (DiffSrev), and Constraint-Based Routing (CBR),
it becomes powerful and complementary tool for Quality of Service (QoS)
handling in IP networks [22]. The MPLS architecture will be described
later.
Figure (2-1): MPLS forwarding label between routers [23].
2.3 MPLS Technology
2.3.1 MPLS Header
When data packets reaches the Label Edge Router (LER), MPLS header is
placed in between layer 2 and 3 of the OSI model. The MPLS header,
which is structured into four parts, has a total length of 32 bits as shown in
Figure (2-2)
• Label (20 bits) acts as an identifier of Forwarding Equivalence
Class (FEC) [15], and also used for determining the Label Switched
Path (LSP).
• Experimental (3 bits) (EXP) indicates Class of Service (CoS).
Chapter Two Concepts of MPLS and QoS
• Bottom of Stack (1 bit): Stack field (S) is used for indicating
whether the label is in the bottom of Stack. If the Label is at the last
entry of stack then the value is set to one else is set to zero.
• Time-to-Live (TTL) (8 bits): This is the last part of the header
where its value decreases by one on every hop as it passes through
the Label Switch Router (LSRs). The packet is dropped when the
TTL value reaches zero [15].
Figure (2-2): MPLS header building block.
2.3.2 MPLS Architecture
The MPLS network is described as “a contiguous set of nodes which
operate MPLS routing and forwarding” [6]. The core consists of Label
Switch Routers (LSRs) and the edge of MPLS domain consists of Label
Edge Routers (LERs).
a. Label Switch Router (LSR)
LSR is a router whose interfaces operate in MPLS backbone. Their main
function is to label the MPLS packets sent to them. When the packet arrives
at the LSR, the LSR checks the look-up table, and then assigns the next
hop. Before forwarding the packet to the next hop, a new label is attached
to the MPLS header after removing the old label [15]. Figure (2-3) shows
the LSR architecture.
Chapter Two Concepts of MPLS and QoS
Figure (2-3): LSR architecture [24].
b. Label Edge Router (LER)
The LER acts as entry point and is called ingress router when the packets
enter into MPLS domain through it. The ingress LER is responsible for
adding the MPLS header to the incoming IP packets [24]. The packets
depart from the MPLS domain by the egress LER router. LER has an
ability to handle layer 3 lookups, and it is in charge of adding or eliminating
the labels from the packets as they enter or exit the MPLS domain [15] as
shown in Figure (2-4)
Figure (2-4): LER architecture [24].
Chapter Two Concepts of MPLS and QoS
c. Forwarding Equivalence Classes (FEC)
An FEC is the class which comprises the group of packets which are treated
in same manner by the LSR [21]. This concept of FEC provides scalability
and flexibility in the network. Flow states should never be managed on
individual flows rather they should be managed on aggregation to ensure
the scalability; for this MPLS uses FEC. The packets belonging to the same
FEC have the same label. But some packets do not belong to the same FEC
and forwarding mechanism due to a different EXP value. Figure (2-5)
shows flow of packet in FEC. Ingress LER decides which packet belongs
to which FEC by examining the IP header of the incoming packets. This is
done only once in MPLS network [25], while in conventional routing the
packets are assigned to FEC within each hop.
Figure (2-5): Flow of packet in FEC.
With the help of packet destination address and the port, the packet
leaves the ingress LER which is then forwarded by the LSRs. It is carried
out by checking the labels and matching them with the ones that are within
LSR’s Label Information Base (LIB) table [26].
d. Label Switched Path (LSP)
The LSP is a path set up by the ingress LER, LSR, and the egress LER to
forward the MPLS labeled packets of a particular FEC. This is done by
using the label swapping forwarding mechanism in such a way that the set
Chapter Two Concepts of MPLS and QoS
up path has the best route [27]. The best route here means the path with
maximum available bandwidth, low jitter, less congestion and low latency.
LSP is the unidirectional path for the label to travel end-to-end through the
MPLS network. There are two kinds of LSPs, control driven LSPs (also
known as hop by hop LSPs) and explicitly routed LSPs (also referred to as
Constraint-based routed LSPs, CR-LSPs) [16], Table (2-1) represents a
simple comparison between these two types of LSPs.
Table (2-1): A basic comparison between Hop-by-Hop and Explicit routing LSPs
Issue
Hop-by-Hop routing
Explicit routing
Topology
route
Everywhere route
Edge route only
Signaling
not required
Requires CR-LDP or RSVP-TE
Routing
Fixed
QoS, Policy, or arbitrary
TE
Difficult
Easy
2.4 MPLS Functionality
The MPLS functionality covers two phases
• The control plane and forwarding (data) plane which are separated
components.
• Label-swapping forwarding mechanism.
a. MPLS Planes
A major concept in MPLS is the differentiation of an IP router’s tasks into
two parts, control plane (control component) and forwarding plane (data
plane) components [28]. Each MPLS plane has two tables. The control
plane has Routing Information Base (RIB) table which contains routing
information from the routing protocols such as Intermediate System-to-
Intermediate System (IS-IS), Open Shortest Path First (OSPF), and Border
Chapter Two Concepts of MPLS and QoS
Gateway Protocol (BGP). The other control plane table is the Label
Information Base (LIB) table which contains the label information that is
obtained from the label bindings that are swapped between the routers.
Because of that the LIB table is smaller than the RIB table; the MPLS
enhances the packet forwarding and makes it faster [26].
The forwarding (data) plane contains two tables too. The
Forwarding Information Base (FIB) table is a routing table which is
derived from the RIB table and contains information about the routes in the
network. This table is used by the ingress LER to decide the next hop to
the unlabeled IP packets, then adds the label to the packet and forwarded
it. The other forwarding (data) plane table is the Label Forwarding
Information Base (LFIB) table which is derived from the LIB and FIB
tables and contains label information related to the best routes. LSR routers
use this table to forward the labeled MPLS packets [29]. Figure (2-6) shows
the control and forwarding (data) components in MPLS routers.
Figure (2-6): Control and Forwarding components in MPLS.
Chapter Two Concepts of MPLS and QoS
b. Label-Swapping Forwarding Mechanism
The label operation is divided into three major parts: push, swap, and pop
operations. The Label Imposition (Push) operation is performed by the
ingress LER during which the classification and labeling of the packets are
done. The Label Swapping/ Switching operation is performed by the LSRs
during which forwarding of packets using labels is done. The Label
Disposition (Pop) operation is performed by the egress LER during which
removal of label is carried out and the original packets are forwarded out
of MPLS network depending on routing table [27]. Figure (2-7) shows the
diagrammatic representation of how these three operations are performed
on the Labels. In Figure (2-7), the IP router connects the customer with the
MPLS network.
Figure (2-7): MPLS Label Operations.
2.5 MPLS Traffic Engineering
Traffic Engineering is the process of choosing network paths by controls
of the traffic flows in the network. Thus the traffic can be balanced over
the different routes, congestions are reduced, and that provides optimum
performance by optimally utilizing the network resources [5]. Traditional
IP network forwards the packets by looking into its destination address at
Chapter Two Concepts of MPLS and QoS
each hop using shortest path metric. The cost of the shortest path is
calculated by using the time that packet takes to reach the next hop. Thus
the shortest link becomes heavily congested when the traffic in the network
increases. In contrast, the links with longer paths are underutilized resulting
in the uneven loads in the links available, at the cost of traffic resources
[9]. However, forwarding traffic through specific path is not achievable
with IP, because the IP forwarding decision at each hop is based on the
destination address, which is often made independently [30]. These
problems are addressed by using Constraint-Based Routing (CBR) control
protocol as a development of MPLS.
CBR algorithm is an extended of shortest path algorithm, also called
Constrained Shortest Path First (CSPF). The constraints that are considered
in CBR algorithm for path computation include, end-to-end delay, the
bandwidth requested, network topology, and other administrative policies.
Calculating a path that meets these constraints needs the information about
the constraints available for each link. This information is forwarded to all
the nodes that are involve in path calculation. This means that the related
link properties have to be flooded throughout the network by including the
TE-specific extensions to the link-state protocols (IS-IS and OSPF). The
CBR procedure for selecting the path involves that, all the routes that have
insufficient bandwidth are removed, also all the routes that do not satisfy
the required constraints are removed. Only the routes that satisfy the
administrative polices are used to choose the path. Then the modified
version of the shortest path algorithm that is utilized by the ingress router
is used to select the path from ingress to egress routers in MPLS domain
[16].
Chapter Two Concepts of MPLS and QoS
There are four primary elements that have to be considered in Traffic
Engineering of a packet network [31]
a. Distribution of topology information: The nodes of a network
must be capable of advertising up to date information about all the
links available and the link failure.
b. Path selection: In this process the shortest paths among all
available links are selected and sometimes when the information
about the bandwidth of the link is available it is also considered in
the process of path selection when using Traffic Engineering.
c. Directing traffic along computed paths: When the path selection
determines the paths considering the constraints such as shortest
path and bandwidth, the nodes must be able to forward the packets
to the defined path, which is done by using forwarding table. This
requires the use of signaling protocols to set up the path.
d. Traffic management: It is responsible for the management of the
traffic in such a way that a user receives his guaranteed QoS. This
may involve flow identification, traffic polishing and traffic
scheduling.
2.6 Signaling Protocols
Signaling protocols are used to set up the paths for the packets to follow,
these paths are commonly known as Label Switched Path (LSP). There are
certain requirements of the signaling protocols that the designed protocols
must follow in order to support the Traffic Engineering feature in MPLS
network. The requirements are listed below [16]
Chapter Two Concepts of MPLS and QoS
a. Robustness: The signaling system must be reliable and should
timely deliver the message in any kind of circumstances such as
congestion and network failure.
b. Scalability: The growth in the size of traffic and the nodes should
not hamper the performance of the signaling system.
c. LSP establishment/teardown/maintenance: Establishing the LSP,
its teardown and its maintenance depending upon the conditions is
another requirement of the signaling system.
d. Specification of QoS: Signaling must be capable of carrying out
trffic characteristics such as bandwidth requirement and also the QoS
requirement.
e. LSP priority/preemption: LSP must make sure that the traffic with
high priority gets through the congested route even if the low priority
traffic has to be torn down.
f. Flexibility in the path setup options: The signaling must be able to
manage the loose and strict CR-LSP setup by the administrator.
g. Alternative path setup and rerouting capability: In case the LSP
is torn down, the signaling must be capable of path optimization,
recovery of the path and of course the maintenance of the LSP path
to provide support for Traffic Engineering.
Chapter Two Concepts of MPLS and QoS
To set up the paths there are many signal protocols which can be used.
In this thesis the two well-known signaling protocols that support Traffic
Engineering are used.
2.6.1 Constraint-based Routing Label Distribution Protocol
Label Distribution Protocol (LDP) has been extended to Constraint-based
Routing Label Distribution Protocol (CR-LDP) to support Constraint-
based Routing Label Switched Paths (CR-LSPs). The constraint means that
for each set of nodes in a network there are a set of limitations that should
be satisfied for the link or links between two nodes to be selected for an
LSP [32]. CR-LDP and LDP protocols don’t require periodic refreshing of
information, it is considered a hard state protocols, that means the signaling
message are sent only once as shown in Figure (2-8).
Figure (2-8): CR-LDP operation.
The four types of LDP messages are [33]
1. Discovery messages: discover and preserve the existence of an LSR in
an MPLS domain. This message is sent as a Hello message periodically
through a UDP port with the multicast address of all routers on this subnet.
2. Session message: is sent to establish, maintain and terminate sessions
between LDP peers.
Chapter Two Concepts of MPLS and QoS
3. Advertisement messages: create, alter, and cancel label mappings for
FECs.
4. Notification messages: indicate the status, diagnosis and error
information. Session, advertisement, and notification messages are
transported over TCP.
To establish an explicit route, a label request message which
includes a listing of nodes over the constraint-based route to be passed is
sent. Then signaling message following the selected path will be sent to the
destination, and if the requested path is capable of meeting the
requirements, the label mapping messages assign and distribute the labels
starting from the destination and transmit the labels in the reverse direction
to go to the source [34]. Supposing that resources are available, the LSP
established is done after a single round-trip of the signaling message. CR-
LDP is able to establish both loose and strict path setups, path
reoptimization, and path preemption [4]. In CR-LDP, the reporting of
failures procedure depends on the TCP layer transport operations of ingress
and egress routers, as shown in Figure (2-9).
Figure (2-9): CR-LDP initiation [19].
CR-LDP enables multiprotocol operations by using FEC that allows
core LSRs to be incurious with regard to the kind of traffic being
transferred over the network. Also for security objectives, the FECs are
Chapter Two Concepts of MPLS and QoS
used, the LSRs are not enabled to find out the transport data services
identity [4].
2.6.2 Resource Reservation Protocol (RSVP)
RSVP is a receiver-oriented protocol, meaning that label allocation and
bandwidth reservation are driven by the receiver node. The RSVP is a soft
state resource reservation, support integrated services focusing on
enterprise networks [4]. RSVP is inherently a soft state protocol, which
uses RESV commands and path to setup an LSP. RSVP depends on
protocol ID and the destination IP address, raw IP datagram routing are
used to transfer packets. A path message is used by the LER to tell each
router along the chosen LSP to acknowledge that this is a desired LSP to
be established. At each node along the path, the RESV messages and the
path are used to refresh the path and reservation states periodically. In
resource reservation, problems can occur because of the RSVP soft state
mechanism along the chosen LSP. However, there is no assurance that the
resources will be reserved depending on the end-to-end request. To support
Explicitly Routed Label-Switched Path (ER-LSP), RSVP-TE has been
proposed and made [35]. A considerable amount of amendments and
expansion have been made to support MPLS LSP setups to keep up with
the traffic engineering requirements. The main amendments and
expansions occur in the areas of resolving scalability problems and traffic
engineering capabilities. Also, the revised RSVP protocol supports both
loose and strict ER-LSPs, also RSVP supports hop-by-hop downstream-on
demand ordered made by using the hop-by hop routing to determine where
to send the PATH message [4].
The following steps depict The CR-LSPs established by means of the
RSVP protocol in MPLS network [4]
Chapter Two Concepts of MPLS and QoS
a. In the MPLS network, the ingress LER chooses a LSP and sends the
Path message to each LSR along this LSP to inform them that this is
the desired LSP used to establish CR-LSP.
b. The LSRs along the chosen LSP reserve the resources and this
information is transmitted to the ingress LER by using the RSVP
message.
c. The Path and RSVP messages are periodically transmitted to all LRs
along the CR-LSP to refresh the state that is maintained as shown in
Figure (2-10).
Figure (2-10): RSVP operation [19].
An RSVP session includes the following objects [36]
• A Path State Block (PSB): maintains the PATH state of the session and
stores the path message.
• A Reservation State Block (RSB): maintains the RESV state of the
session and stores the RESV message.
• MPLS Labels: are stored / reserved for the session to use to Traffic.
• RSVP is the soft state protocol: It uses Path and RSVP commands to
establish a path.
Chapter Two Concepts of MPLS and QoS
2.7 Quality of Service
QoS is defined as a set of techniques to classify and manage network
resources with the help of which a certain level of packet loss, bit rate,
jitter, delay, etc., can be guaranteed [37]. The main objective of QoS is to
assure a guaranteed service and support as a framework to the Internet. The
IETF suggested several service models, mechanisms, policies and schemes
for satisfying QoS demands.
2.7.1 Quality of Service Models
There are three QoS models including Best Effort (BE), Integrated Service
(IntServ), and Differentiated Service (DiffServ) model. These models are
described below
a. Best Effort (BE) Service Model: This is a traditional datagram
service model. There is no QoS in this model because there is no
assurance of reliability, throughput, and delay in it. BE has a single
queue and the First-In First-Out (FIFO) queue is used as default for
scheduling [38].
b. Integrated Service Model (IntServ): IntServ uses RSVP protocol
to accomplish the path reservation. It is multiple service model
which can take different QoS requirements under consideration. It
can provide end-to-end service guarantees in connectionless IP
networks. The main problem with the IntServ architecture is the
scalability issue, it must be configured on each router along the path
and requires a huge storage and processing overhead on the routers.
It is can do well in small scale networks [10].
Chapter Two Concepts of MPLS and QoS
c. Differentiated Service (DiffServ): This model came into existence
for overcoming the underlying problem of scalability in IntServ. It
provides differing levels of QoS to different traffic flows. It cannot
provide per flow bandwidth and delay guarantees, but it makes the
stateless network scalable and robust [39]. In a DiffServ domain, all
IP packets crossing a link and requiring the same DiffServ behavior
are said to constitute a Behavior Aggregate (BA). At the ingress
node of the DiffServ domain packets are classified and marked with
a DiffServ Code Point (DSCP) which corresponds to their BA.
Epach BA is identified by a single DSCP and sending packet
according to its priority. At each transit node, the DSCP is used to
select the Per-Hop Behavior (PHB) that determines the scheduling
(queuing, dropping) treatment [10]. The packets which are specified
or marked need to be forwarded on a PHB basis as quickly as
possible at the network boundaries based on buffer management and
packet scheduling mechanisms indicated by the DSCP field value
[40].
In IPv4, the DSCP field defines the layout of the Type of Service
(ToS) octet. For special treatment of the delivering packet, PHBs are vitally
important in packet forwarding path along with traffic conditioning
mechanisms needed to be done by using some monitoring, marking,
metering policing and shaping at network nodes. The DiffServ is identified
in the packet forwarding path by mapping the DSCP to a particular
forwarding treatment at each network node along its path. PHBs are
deployed for a range of queue service disciplines, for instance, Weighted
Fair Queue (WFQ) or Priority Queue (PQ) management on node’s output
interface queue of a network. Figure (2-11) shows an example of
aggregation where the flows are aggregated based on their DSCP code
Chapter Two Concepts of MPLS and QoS
(yellow or red) and single user information (A, B, C or D) experiences a
predefined PHB, e.g. EF (Expedited Forwarding), AF (Assured
Forwarding), and BE (Best Effort) after passing the DiffServ router [41].
Figure (2-11): A DiffServ aggregation behavior [41].
Figure (2-12) shows the architecture of DiffServ domain. The
ingress and egress routers work as a downstream or upstream node of a
boundary link in a given traffic direction. The ingress edge router classifies,
measures and marks the packet appropriately to be selected a PHB from
one of the PHB groups within the domain [42]. The core routers put the
packets in queues and treat them according to their priority by using one of
the queue scheduling mechanisms.
Figure (2-12): DiffServ Architecture [38].
Chapter Two Concepts of MPLS and QoS
2.7.2 Classification and Conditioning of Traffic in DiffServ
The traffic classification and conditioning represent important steps which
perform marking, policing, shaping, and metering for confirming the traffic
entry to the DiffServ domain. This whole process should be completed
satisfying the rules mentioned in the Traffic Conditioning Agreement
(TCA) and Service Level Agreement (SLA) [38].
a. Classification: The classifier selects packets from the traffic stream
depending on the DSCP value in the IP header. The classification is
done only once at the ingress edge router of the DiffServ domain.
b. Policing: Policing is the process in which the packets are discarded
within the traffic stream so as to comply with the rules mentioned in
SLA.
c. Shaping: Shaping is the process in which the packets are delayed so
as to comply with the rules mentioned in SLA.
d. Marking: Marking is the process in which the DSCP value is set in
accordance with the set of defined rules like remarking, pre-
marking.
e. Metering: Metering is the process in which the classifier selects the
traffic stream and measures the temporal properties like rate of those
selected traffic stream.
2.8 Queue Scheduling Mechanisms
Network traffic has different traffic characteristics and may have different
QoS requirements due to the presence of voice, video, data, etc., in the
Chapter Two Concepts of MPLS and QoS
same network. So a good scheduling mechanisms deal with the various
flows in a fair and suitable manner designed in order to improve the quality
of service [43]. In this thesis three types of queuing algorithms are used
FIFO queuing, Priority Queuing (PQ), and Weighted Fair Queuing (WFQ).
a. First-In First-Out (FIFO) Queuing
This is a Best Effort (BE) service model in which all packets are treated
equally by placing them in a single queue. The idea of FIFO queuing is that
the first packet that arrives at a router is the first packet to be transmitted.
Packets wait in a queue until the router is ready to process them. By default
the maximum FIFO queue size is 500 packets. If a packet arrives and the
queue is full, then the router discards (drops) that packet. This is done
without regard to how important the packet is [44]. Figure (2-13) shows a
conceptual view of a FIFO queue
Figure (2-13): FIFO Queue concept [43].
b. Priority Queuing (PQ)
This technique is categorized under DiffServ service model. In priority
queuing, packets are first allocated to a priority class. Each priority class
has a queue assigned to it. Principle idea of Priority Queuing depends on
the priority of the packets, a highest priority is transmitted on the output
port first and then the packets with lower priority and so on. Priority
Queuing has four preconfigured queues high, medium, normal and low
Chapter Two Concepts of MPLS and QoS
priority queues. By default, the maximum sizes of these queues are 80, 60,
40 and 20 packets capacity, respectively [45]. The system does not stop
serving a queue unless the queue becomes empty [43], this means that the
highest priority traffic experiences minimal delay, while traffic with lower
priority levels might suffer the problem of resource starvation when those
with highest priority remain occupying the queue. So, the main problem of
this algorithm is the resource management for traffic with the lower priority
[45]. A simple example of the priority queuing with two levels is shown in
Figure (2-14)
Figure (2-14): Priority queuing concept [43].
c. Weighted Fair Queuing (WFQ)
This approach is also classified as DiffServ model. It is a flow-based
queuing algorithm which supports flows with different bandwidth
requirements [46]. It doesn't allow classification options to be configured.
Based on flows, WFQ classifies packets automatically, with each flow
being placed into a separate queue. By default, the maximum queue size is
500 packets capacity. In WFQ, a flow means all packets with the same
source/destination IP address, transport layer protocol, TCP or UDP
source/destination port. Because WFQ puts packets of different flows in
Chapter Two Concepts of MPLS and QoS
different queues, it must have a greater number of queues than all of the
non-flow-based queuing algorithms [47]. In this technique, the packets are
still allocated to various classes and admitted to different queues. However,
the queues are weighted based on the priority of the queues; higher weight
means a higher priority. The system uses a round-robin style to process
packets in each queue. The number of selected packets from each queue is
based on the corresponding weight. For instance, if the weights are 3, 2,
and 1, three packets are processed from the first queue, two from the second
queue, and one from the third queue [43]. If all flows have the same
priority/weight, WFQ effectively divides the interface bandwidth and
distributes the bandwidth fairly among all the existing flows. Therefore,
low-volume interactive flows are scheduled and do not end up with packets
waiting in their corresponding queues. High-volume interactive flows
build up their corresponding queues and end up with packets waiting,
resulting in more delay and possibility to drop packets. Figure (2-15)
shows the technique with three classes.
Figure (2-15): Weighted fair queuing concept [43].
Chapter Two Concepts of MPLS and QoS
2.9 Performance Metrics
The parameters that are adopted for the study are
a. Number of Maintained Calls
The number of maintained calls is calculated as follows [19]
Number of maintained calls = (drop time – start time) /2 (2.1)
where
Drop time = the time at which traffic begins to drop packets.
Start time = time of start calls in the network.
At the 10th second of the simulation, the first VoIP call is formed. As this
time is used to train the network for the current environment, then for each
2 seconds, a call will be formed.
b. Voice Jitter
Jitter is defined as the signed maximum difference in one-way delay of the
packets over a particular time interval. Let t(i) and t'(i) be the time of
transmitted a packet at the transmitter and the time of received the same
packet at the receiver, respectively. Jitter is then calculated as [48]
According to Equation (2.2), the jitter value can be negative which means
that the time difference between the packets at the destination is less than
that at the source.
Chapter Two Concepts of MPLS and QoS
c. Packet Delay Variation (PDV)
Packet delay variation (PDV) plays a crucial role in the network
performance degradation and affects the user-perceptual quality. Higher
packet delay variation results in network congestion of the packets. PDV
is defined as the variance of the packet delay which can be calculated as
[48]
(2.3)
where is the average delay of the n selected packets.
d. Packet End to End Delay (De2e)
The total voice packet delay, De2e is estimated from the following formula
where is the network delay, is the encoding delay, is the decoding
delay, is the compression delay, and is the decompression delay [49].
In order to establish acceptable call for VoIP applications, it is
required that end-to-end packet delay should not override 150 millisecond.
The voice delay in G.711 can be divided into two contributing components,
which are described as follows [50]
i. At the transmitter G.711 codec introduces 1 millisecond for
encoding and 20 millisecond for packetization. Therefore, the
delay is approximated to a fixed delay of 25 millisecond
considering these two delays along with compression.
Chapter Two Concepts of MPLS and QoS
ii. The delay introduced at the receiver is coming from buffering,
decompression, depacketization and playback delay. The total
delay due to the above factors is approximated to a fixed delay of
45 millisecond. The maximum acceptable network delay can be
calculated from the above transmitter and receiver delays as
150-25-45 = 80 millisecond. The 150 millisecond represents the
maximum acceptable end-to-end delay so that the quality of the
established VoIP call is acceptable.
e. Mean Opinion Score [51]
MOS provides a numerical measure of the quality of human speech in
voice telecommunications, with value ranging from 1 to 5 where 1 is the
worst quality and 5 is the best quality. The MOS value depends on R factor
where
R
= Effect of impairments that occurs with the voice signal.
= Impairments caused by different types of losses occurring due to
codec's and network.
= Impairment caused by delay particularly mouth-to-ear delay.
A= Constant.
Using the default settings for, Equation (2.5) reduces to
Chapter Three Performance Evaluation of MPLS TE Signal Protocols for Voice Applications
Chapter Three
Performance Evaluation of MPLS TE Signal Protocols for
Voice Applications
3.1 Introduction
This chapter studies the performance of MPLS-TE networks for VoIP
applications. Two signal protocols are considered, namely Constraint
based Routed-Label Distribution Protocol (CR-LDP) and Resource
Reservation Protocol (RSVP). Further different audio codecs including
PCM, GSM FR, G.723.1, G.726, G.728, G.729a and IS-641 are used in the
investigation (see Appendix A). The employed simulation environment is
based on OPNET 14.5 simulator.
3.2 Network Setup
Two different MPLS networks are implemented as shown in Figures
(3-1) and (3-2), each one is simulated with CR-LDP and RSVP signal
protocols. The first network consists of six routers and four switches and
the second network consists of eight routers and four switches. The routers
in each network are connected through DS3 cable with data rate of 44.736
Mbps. The end nodes are connected to the network via switches. Links of
each switch are 100BaseT. To simulate real network environments, different
applications are used in each network such as voice, video, HTTP, FTP, DB,
Telnet and Email. This study focuses on voice traffic, so the other applications
are inserted as secondary traffic. The VoIP traffic is sent from source (Voice 1)
to destination (Voice 2), the video traffic is sent from source (Video 1) to
destination (Video 2), DB and HTTP traffic is sent from source (DB, HTTP) to
destination (DB, HTTP server), remote traffic is sent from source (remote) and
FTP traffic is sent from source (FTP) to destination (FTP, remote server).
Chapter Three Performance Evaluation of MPLS TE Signal Protocols for Voice Applications
Figure (3-1): First MPLS network topology.
Figure (3-2): Second MPLS network topology.
Chapter Three Performance Evaluation of MPLS TE Signal Protocols for Voice Applications
The voice workstations use different types of audio codecs, namely, PCM
(64 Kbps), GSM FR (13 Kbps), G.723.1 (5.3 Kbps), G.726 (16 Kbps),
G.728 (16 Kbps), G.729a (8 Kbps), and IS-641 (7.4 Kbps) each type of
codecs is simulated individually.
3.3 Simulation Parameters
Table (3-1) defines the parameters of the applications to be used
throughout the simulations. FTP work stations use files of size 5000 byte.
HTTP work stations use pages of size 1000 bytes. HTTP/1.1 is a revision
of the HTTP (HTTP/1.0). In HTTP/1.0, a detached connection to the same
server for each resource request is made. HTTP/1.1 can reuse a connection
multiple times to download images, page inter-arrival times are
exponentially distributed with mean 60 sec, that is employed in this study.
Emails are sent with inter-arrival times exponentially distributed with
mean equal 360 sec. For database application, the transactions arrive with
inter-arrival times exponentially distributed with mean equal 30 sec. Telnet
application initiates commands from terminal to telnet host, these
commands consist of a normally distributed number of bytes with mean
equal 10, variance equal 4. Best-effort service means that the user obtains
unspecified variable bit rate and delivery time, depending on the current
traffic load of the network. Video workstations transfer 10 frames per sec;
each frame consists of 128120 pixels. Voice workstations are applied
with seven types of audio codecs.
Chapter Three Performance Evaluation of MPLS TE Signal Protocols for Voice Applications
Table (3-1): Application parameters.
FTP Applications
File size
5000 bytes
Type of Service
Best effort
HTTP Applications
HTTP Specification
HTTP1.1
Page Properties
1000 bytes
Type of Service
Best effort
E-Mail Applications
E-Mail Size
2000 bytes
Type of Service
Standard
Data base Applications
Transaction Size
16 bytes
Type of Service
Best effort
Print Applications
File size
3000-90000 bytes
Type of Service
Best effort
Telnet Applications
Terminal Traffic
Mean =10, variance = 60 bytes
Type of Service
Best effort
Video Applications
Frame rate
10 frames/sec
Frame Size
128120 pixels
Type of Service
Streaming Multimedia
Voice Applications
Encoder Scheme
7 codecs types are used
Type of Service
Interactive Voice
Chapter Three Performance Evaluation of MPLS TE Signal Protocols for Voice Applications
3.4 Simulation Results
The results are presented as comparison between the performance of
CR-LDP and RSVP signal protocols and displayed in series of figures. In
each figure, the green line represents the sent voice packets/sec (packet =
80 bytes) in which the blue and red lines represent the received packets/sec
when CR-LDP and RSVP signal protocols are used, respectively. Each
figure consists of five parts describing the variation of the following
measure parameters with time
a. The first part shows the sent and received voice packets/sec. In
this part, the traffic drop time (which is the point at which the
received line diverts from the sent line) is used to calculate the
number of maintained calls according to Equation (2.1).
b. The other parts of this figure show the results of jitter, PDV, and
end-to-end delay versus simulation time. It is observed that at
traffic drop time, the values of jitter, PDV, and end-to-end delay
begin to increase significantly.
Figure (3-3) is presented as an example which is related to G.723.1
audio codec in the 2nd network. The first part of this figure shows the voice
packets start to drop from 460th sec of the simulation time when using CR-
LDP protocol as indicated by point 1, while when using RSVP protocol the
drop of packets start from 40th sec as indicated by point 2. After the packet
is dropped, the quality of VoIP-calls-will-be-unacceptable due to the-
packet loss which-causes voice breaks-and-voice-skips. The second part in
the same figure shows the voice packet jitter. It is noted that voice jitter
begins to increment at the traffic drop time to be 0.041 millisecond in CR-
LDP as indicated by point 3, and reaches 2.698 millisecond in RSVP as
Chapter Three Performance Evaluation of MPLS TE Signal Protocols for Voice Applications
indicated by point 4. The third part of the figure shows the voice PDV
where point 5 represents the PDV value at the traffic drop time for CR-
LDP which is 0.003 msec. For RSVP, it is 0.222 millisecond as indicated
by point 6. The fourth part shows the end-to-end delay, which equals
111.531 millisecond for CR-LDP as indicated by point 7 and 123.505
millisecond for RSVP protocol as indicated in point 8. The last part of the
figure is for MOS and shows that the quality of speech is affected by the
drop of packets and delays. Thus MOS decreases in this case and is equal
2.546 this means that quality of speech is between the good and fair
conditions as indicated by points 9 and 10. All results are put in the same
manner as that shown in Figure (3-3). The results in Figure (3-3) are related
to G.723.1 in the 2nd network.
Figures (3-4) through (3-17) show the simulation results related to
both networks when operating with different types of codecs. The
maximum values of the performance metrics for the interval between the
beginnings of the simulation and the traffic drop time are recorded in the
Table (3-2).
Chapter Three Performance Evaluation of MPLS TE Signal Protocols for Voice Applications
Figure (3-3): Example of results presentation.
.
Chapter Three Performance Evaluation of MPLS TE Signal Protocols for Voice Applications
Figure (3-4): Scenario PCM codec in the 1st network.
Figure (3-6): Scenario GSM FR codec in the 1st
network.
Figure (3-5): Scenario PCM codec in the 2nd network.
Figure (3-7): Scenario GSM FR codec in the 2nd
network.
Chapter Three Performance Evaluation of MPLS TE Signal Protocols for Voice Applications
Figure (3-8): Scenario G.723.1 codec in the 1st network.
Figure (3-10): Scenario G.726 codec in the 1st network.
Figure (3-9): Scenario G.723.1 codec in the 2nd network.
Figure (3-11): Scenario G.726 codec in the 2nd network.
Chapter Three Performance Evaluation of MPLS TE Signal Protocols for Voice Applications
Figure (3-12): Scenario G.728 codec in the 1st network.
Figure (3-14): Scenario G.729 codec in the 1st network.
Figure (3-13): Scenario G.728 codec in the 2nd network.
Figure (3-15): Scenario G.729 codec in the 2nd network.
Chapter Three Performance Evaluation of MPLS TE Signal Protocols for Voice Applications
Figure (3-16): Scenario IS-641 codec in the 1st network.
Figure (3-17): Scenario IS-641 codec in the 2nd network.
Chapter Three Performance Evaluation of MPLS TE Signal Protocols for Voice Applications
Table (3-2): Performance Metrics for CR-LDP and RSVP Signal Protocols-based networks.
Codec
types
Network
number
Traffic drop
time (second)
No. of
calls/second
Jitter
(millisecond)
PDV
(millisecond)
End-to-end delay
(millisecond)
MOS
CRLDP
RSVP
CRLDP
RSVP
CRLDP
RSVP
CRLDP
RSVP
CRLDP
RSVP
CRLDP
RSVP
PCM
1
100
38
45
14
0.0020
0.0947
0.0030
0.0660
79.467
76.834
3.595
3.595
2
125
40
57
15
0.0461
0.2061
0.0200
0.0493
78.400
87.500
3.674
3.674
GSM
FR
1
102
40
46
15
1.7600
1.6100
0.4080
0.0670
126.400
114.500
3.467
3.467
2
318
40
154
15
0.2930
1.9630
0.5940
0.2900
142.964
124.584
3.548
3.548
G.723.1
1
460
102
225*
46
0.2120
4.1170
0.0370
0.4600
148.105
137.048
2.559
2.559
2
460
40
225*
15
0.0410
2.6980
0.0030
0.2220
111.531
123.505
2.546
2.546
G.726
1
166
46
78
18
0.1930
0.1278
0.1900
1.8100
111.500
112.300
3.595
3.595
2
162
42
76
16
0.0340
0.3929
0.0060
0.1870
70.800
128.800
3.674
3.674
G.728
1
166
45
78
17
0.1730
0.2050
0.1710
0.9330
113.900
141.138
3.562
3.562
2
160
40
75
15
0.0180
0.3496
0.0020
0.5940
70.900
81.100
3.675
3.675
G.729a
1
160
40
75
15
0.0370
0.3990
0.0023
0.3190
81.200
98.600
3.000
3.000
2
166
40
78
15
0.0230
0.3462
0.1100
0.1700
127.800
106.300
3.059
3.059
IS-641
1
310
30
150
10
0.0050
0.3830
0.0200
0.2300
109.300
103.700
3.468
3.468
2
316
40
153
15
0.1330
1.1320
0.0070
0.1410
104.046
116.862
3.550
3.550
Best results
Chapter Three Performance Evaluation of MPLS TE Signal Protocols for Voice Applications
3.5 Result Summary
a. Number of Maintained Calls
From the previous results it is noted that the number of maintained calls
when using CR-LDP is larger than those with using RSVP in all cases.
Figure (3-18) clarifies the difference between the performances of these
two signal protocols.
Figure (3-18): Number of maintained calls/sec in both networks.
b. Voice Jitter
Except in the GSM FR and G.726 codecs used in the first network, it is
observed that CR-LDP signal protocol has lower jitter values than RSVP.
The highest jitter is obtained when G.723.1 codec is used with RSVP
protocol and this value decreases to 5.14% in the first network and to
1.51% in the second when CR-LDP protocol is used as shown in Figure
(3-19)
45 46
225
78 78 75
150
14 15
46
18 17 15 10
57
154
225
76 75 78
153
15 15 15 16 15 15 15
0
50
100
150
200
250
PCM GSM G.723.1 G.726 G.728 G.729A IS-641
no. of calls /sec
No. of calls for both networks
CR-LDP_NET1 RSVP_NET1 CR-LDP_NET2 RSVP_NET2
Chapter Three Performance Evaluation of MPLS TE Signal Protocols for Voice Applications
Figure (3- 19): Voice jitter for both networks.
c. Packet Delay Variation
Except the GSM FR codec, it is observed that CR-LDP has lower values
than RSVP. The highest PDV value is obtained when G.726 codec is used
with RSVP in the first network and this value decreases to 10.49% when
CR-LDP is used, see Figure (3-20)
Figure (3-20): Voice PDV for both networks.
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
PCM GSM G.723.1 G.726 G.728 G.729A IS-641
Time (msec)
jitter for both networks
CR-LDP_NET1 RSVP_NET1 CR-LDP_NET2 RSVP_NET2
0
0.5
1
1.5
2
PCM GSM G.723.1 G.726 G.728 G.729A IS-641
Time (msec)
PDV for both networks
CR-LDP_NET1 RSVP_NET1 CR-LDP_NET2 RSVP_NET2
Chapter Three Performance Evaluation of MPLS TE Signal Protocols for Voice Applications
d. End-to-end Delay
Comparison among codecs shows that G.723.1 has the highest values of
end-to-end delay, while PCM has the lowest values of it as shown in Figure
(3-21).
Figure (3-21): Voice end-to-end delay for both networks.
e. Main Opinion Score
The MOS values shown in Figure (3-22) indicate that the highest MOS
values are obtained when one uses PCM, G.726, and G.728. Slightly less
values are obtained for GSM FR and IS-641. The lowest value is for
G.723.1 preceded by G.29. Investigating the results in Figure (3-22)
reveals that the MOS values are not affected by TE signal protocols.
Figure (3-22): MOS performance in both networks.
0
50
100
150
200
PCM GSM G.723.1 G.726 G.728 G.729A IS-641
Time (msec)
End-to-end delay for both networks
CR-LDP_NET1 RSVP_NET1 CR-LDP_NET2 RSVP_NET2
0
1
2
3
4
PCM GSM G.723.1 G.726 G.728 G.729A IS-641
MOS Values for both networks
CR-LDP_NET1 RSVP_NET1 CR-LDP_NET2 RSVP_NET2
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Chapter Four
Effect of QoS Algorithms on the Performance of MPLS-TE
Networks
4.1 Introduction
This chapter studies the performance of Multiprotocol Label
Switching (MPLS) networks for VoIP applications when applying both
Traffic Engineering (TE) signal protocols and Quality of Service (QoS)
algorithms. The used QoS algorithms are First-In First-Out FIFO), Priority
Queuing (PQ), and Weighted Fair Queuing (WFQ). The study addresses
different audio codecs including PCM (64 Kbps), GSM FR (13 Kbps),
G.723.1 (5.3 Kbps), G.726 (16 Kbps), G.728 (16 Kbps), G.729 (8 Kbps)
and IS-641(7.4 Kbps). Results related to both CR-LDP and RSVP traffic
engineering signal protocols are evaluated and compared in the presence
of different QoS algorithms. The results represent for the two networks
addressed in chapter three.
4.2 Results Representation
4.2.1 First-In First-Out (FIFO) Algorithm
Figures (4-1) through (4-14) represent the results when applying FIFO
algorithm. It is observed that the CR-LDP signal protocol is better than
RSVP with all audio codecs used. This is because when using RSVP
protocol, the packets begin to drop earlier than when using CR-LDP
protocol. This is true except for one case when applying G.723.1 codec in
the first network as shown in Figure (4-5). The implementation of FIFO
algorithm makes the performance worse, but except two abnormal cases
when using (GSM FR with CR-LDP) and (G.723.1 with RSVP) in the first
network as shown in Figures (4-3) and (4-5), respectively.
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Figure (4-1): Scenario PCM codec with applying FIFO
algorithm in the 1st network.
Figure (4-2): Scenario PCM codec with applying FIFO
algorithm in the 2nd network.
Figure (4-3): Scenario GSM FR codec with applying
FIFO algorithm in the 1st network.
Figure (4-4): Scenario GSM FR codec with applying
FIFO algorithm in the 2nd network.
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Figure (4-5): Scenario G.723.1 codec with applying FIFO
algorithm in the 1st network.
Figure (4-7): Scenario G.726 codec with applying FIFO
algorithm in the 1st network.
Figure (4-6): Scenario G.723.1 codec with applying FIFO
algorithm in the 2nd network.
Figure (4-8): Scenario G.726 codec with applying FIFO
algorithm in the 2nd network
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Figure (4-9): Scenario G.728 codec with applying FIFO
algorithm in the 1st network.
Figure (4-11): Scenario G.729 codec with applying FIFO
algorithm in the 1st network.
Figure (4-10): Scenario G.728 codec with applying FIFO
algorithm in the 2nd network.
Figure (4-12): Scenario G.729 codec with applying FIFO
algorithm in the 2nd network.
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Figure (4-13): Scenario IS-641 codec with applying FIFO
algorithm in the 1st network.
Figure (4-14): Scenario IS-641 codec with applying FIFO
algorithm in the 2nd network.
4.2.2 Priority Queuing (PQ)
Figures (4-15) through (4-28) represent the results when applying PQ
algorithm. These results also show that the CR-LDP signal protocol is
better than the RSVP. However, the performance of the CR-LDP decreases
compared to its performance before applying QoS algorithms in term of
number of maintained calls. The less values in terms of voice jitter, voice
PDV, and end-to-end delay are gained by using this algorithm whether
using CR-LDP or RSVP signal protocols.
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Figure (4-15): Scenario PCM codec with applying PQ
algorithm in the 1st network.
Figure (4-17): Scenario GSM FR codec with applying PQ
algorithm in the 1st network.
Figure (4-16): Scenario PCM codec with applying PQ
algorithm in the 2nd network.
Figure (4-18): Scenario GSM FR codec with applying PQ
algorithm in the 2nd network.
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Figure (4-19): Scenario G.723.1 codec with applying PQ
algorithm in the 1st network.
Figure (4-21): Scenario G.726 codec with applying PQ
algorithm in the 1st network.
Figure (4-20): Scenario G.723.1 codec with applying PQ
algorithm in the 2nd network.
Figure (4-22): Scenario G.726 codec with applying PQ
algorithm in the 2nd network
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Figure (4-23): Scenario G.728 codec with applying PQ
algorithm in the 1st network.
Figure (4-25): Scenario G.729 codec with applying PQ
algorithm in the 1st network.
Figure (4-24): Scenario G.728 codec with applying PQ
algorithm in the 2nd network.
Figure (4-26): Scenario G.729 codec with applying PQ
algorithm in the 2nd network.
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Figure (4-27): Scenario IS-641 codec with applying PQ
algorithm in the 1st network.
Figure (4-28): Scenario IS-641 codec with applying PQ
algorithm in the 2nd network.
4.2.3 Weighted Fair Queuing (WFQ) Algorithm
Figures (4-29) through (4-42) represent the results when applying WFQ
algorithm. Investigating these results reveals that, contrary to the previous
results, the RSVP is better than CR-LDP in all used audio codec types
except one case when using GSM FR codec in the second network as
shown in Figure (4-32).
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Figure (4-29): Scenario PCM codec with applying WFQ
QoS algorithm in the 1st network.
Figure (4-31): Scenario GSM FR codec with applying
WFQ algorithm in the 1st network.
Figure (4-30): Scenario PCM codec with applying WFQ
algorithm in the 2nd network.
Figure (4-32): Scenario GSM FR codec with applying
WFQ algorithm in the 2nd network.
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Figure (4-33): Scenario G.723.1 codec with applying
WFQ algorithm in the 1st network.
Figure (4-35): Scenario G.726 codec with applying WFQ
algorithm in the 1st network.
Figure (4-34): Scenario G.723.1 codec with applying
WFQ algorithm in the 2nd network.
Figure (4-36): Scenario G.726 codec with applying WFQ
algorithm in the 2nd network.
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Figure (4-37): Scenario G.728 codec with applying WFQ
algorithm in the 1st network.
Figure (4-39): Scenario G.729 codec with applying WFQ
algorithm in the 1st network.
Figure (4-38): Scenario G.728 codec with applying WFQ
algorithm in the 2nd network.
Figure (4-40): Scenario G.729 codec with applying WFQ
algorithm in the 2nd network.
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Figure (4-41): Scenario IS-641 codec with applying WFQ
algorithm in the 1st network.
Figure (4-42): Scenario IS-641 codec with applying WFQ
algorithm in the 2nd network.
4.3 Results
All maximum values from the beginning of the simulation time to the
traffic drop time are listed in tables as follows
• Table (4-1) : Traffic drop time
• Table (4-2) : Number of maintained calls/sec
• Table (4-3) : Voice Jitter
• Table (4-4) : Voice Packet Delay Variations
• Table (4-5) : End-to-end Delay
• Table (4-6) : Main Opinion Score
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Table (4-1): Traffic drop time (sec).
Codecs
types
Network
number
CRLDP
CRLDP
FIFO
CRLDP
PQ
CRLDP
WFQ
RSVP
RSVP
FIFO
RSVP
PQ
RSVP
WFQ
PCM
1
100
40
55
192
38
30
52
192
2
125
52
54
195
40
30
54
192
GSM FR
1
102
427
175
64
40
32
120
420
2
318
54
170
55
40
32
90
34
G.723.1
1
460
94
495
96
102
602
495
* 600
2
460
56
140
540
40
30
140
* 615
G.726
1
166
55
172
52
46
32
52
216
2
162
55
54
202
42
30
54
216
G.728
1
166
55
170
52
45
34
157
210
2
160
55
54
200
40
32
54
216
G.729
1
160
56
55
192
40
36
52
210
2
166
56
55
198
40
30
55
217
IS-641
1
310
55
55
385
30
33
55
420
2
316
56
55
390
40
33
55
414
Table (4-2): Number of calls/sec.
Types of
codecs
Network
number
CRLDP
CRLDP
FIFO
CRLDP
PQ
CRLDP
WFQ
RSVP
RSVP
FIFO
RSVP
PQ
RSVP
WFQ
PCM
1
45
15
22
91
14
10
21
91
2
57
21
22
92
15
10
22
91
GSM
FR
1
46
208
82
27
15
11
55
205
2
154
22
80
22
15
11
40
12
G.723.1
1
225
42
242
43
46
296
242
* 295
2
225
23
65
265
15
10
65
* 302
G.726
1
78
22
81
21
18
11
21
103
2
76
22
22
96
16
10
22
103
G.728
1
78
22
80
21
17
12
73
100
2
75
22
22
95
15
11
22
103
G.729
1
75
23
15
91
15
13
21
100
2
78
23
22
94
15
10
22
103
IS-641
1
150
22
15
187
10
11
22
205
2
153
23
22
190
15
11
22
202
*Best results
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Table (4-3): Voice Jitter.
Codecs
types
Network
number
CR
LDP
(millisecond)
CRLDP
FIFO
(millisecond)
CRLDP
PQ
(microsecond)
CRLDP
WFQ
(millisecond)
RSVP
(millisecond)
RSVP
FIFO
(millisecond)
RSVP
PQ
(microsecond)
RSVP
WFQ
(millisecond)
PCM
1
0.002
0.004
-0.564
0.027
0.094
0.069
-0.222
0.027
2
0.046
0.038
-0.344
0.041
0.206
0.446
-0.304
0.027
GSM
FR
1
1.760
0.088
-3.180
0.456
1.610
0.664
-6.590
0.000
2
0.293
0.371
0.330
0.003
1.963
0.731
-1.171
1.108
G.723.1
1
0.212
-0.137
0.765
-0.187
4.117
0.066
0.141
0.140
2
0.041
0.018
0.109
0.027
2.698
0.584
0.228
0.141
G.726
1
0.193
0.129
-4.746
1.808
0.127
0.281
-3.554
0.164
2
0.034
0.011
-0.461
0.284
0.392
0.309
-0.447
0.103
G.728
1
0.173
0.162
1.986
1.839
0.205
0.110
-6.185
0.094
2
0.018
0.011
-0.461
0.149
0.349
0.234
-0.653
0.157
G.729
1
0.037
0.068
1.018
0.008
0.399
0.125
-0.306
0.107
2
0.023
0.210
0.499
0.074
0.346
0.008
-0.345
0.191
IS-641
1
0.005
0.101
-0.925
1.786
0.383
0.216
0.319
0.221
2
0.133
0.536
0.511
1.633
1.132
0.501
0.385
0.019
Table (4-4): Packet Delay Variation.
Codecs
types
Network
number
CR
LDP
(millisecond)
CRLDP
FIFO
(millisecond)
CRLDP
PQ
(microsecond)
CRLDP
WFQ
(millisecond)
RSVP
(millisecond)
RSVP
FIFO
(millisecond)
RSVP
PQ
(microsecond)
RSVP
WFQ
(millisecond)
PCM
1
0.003
0.014
0.142
0.000
0.0660
0.092
0.019
0.000
2
0.020
0.027
0.206
0.010
0.0493
0.185
0.013
0.000
GSM
FR
1
0.408
0.007
19.600
3.800
0.0670
0.099
19.600
0.000
2
0.594
0.095
10.300
0.030
0.2900
0.216
33.300
0.000
G.723.1
1
0.037
0.100
11.420
0.130
0.4600
0.000
0.091
0.000
2
0.003
0.020
0.185
0.000
0.2220
0.000
0.247
0.010
G.726
1
0.190
0.037
8.927
0.140
1.8100
0.190
0.168
0.020
2
0.006
0.033
0.215
0.000
0.1870
0.181
0.035
0.000
G.728
1
0.171
0.048
10.270
0.080
0.9330
0.244
16.040
0.010
2
0.002
0.033
0.215
0.010
0.5940
0.098
0.087
0.010
G.729
1
0.002
0.030
0.147
0.000
0.3190
0.351
0.011
0.007
2
0.110
0.039
0.202
0.006
0.1700
0.001
0.033
0.026
IS-641
1
0.020
0.019
0.191
0.007
0.2300
0.114
0.024
0.037
2
0.007
0.021
0.175
0.183
0.1410
0.273
0.027
0.002
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Table (4-5): End-to-end Delay (millisecond).
Codecs
types
Network
number
CRLDP
CRLDP
FIFO
CRLDP
PQ
CRLDP
WFQ
RSVP
RSVP
FIFO
RSVP
PQ
RSVP
WFQ
PCM
1
79.467
81.755
76.399
83.651
76.834
79.086
75.063
83.651
2
78.400
75.800
63.610
79.591
87.500
85.700
63.551
73.033
GSM FR
1
126.400
110.778
97.091
148.30
114.500
113.80
97.790
103.94
2
142.964
117.100
85.360
110.40
124.584
108.40
87.630
139.40
G.723.1
1
148.105
126.100
118.31
139.70
137.048
121.55
117.18
129.05
2
111.531
112.100
103.53
107.59
123.505
138.80
103.89
124.05
G.726
1
111.500
97.500
78.557
98.100
112.300
101.50
76.537
114.63
2
70.800
82.800
63.540
144.40
128.800
89.600
63.467
85.071
G.728
1
113.900
104.400
84.251
100.30
141.138
112.60
78.020
96.100
2
70.900
82.800
63.540
90.538
81.100
77.200
63.399
92.925
G.729
1
81.200
93.100
76.363
80.163
98.600
122.71
80.196
101.42
2
127.800
88.100
63.515
82.109
106.300
64.900
63.435
103.94
IS-641
1
109.300
105.500
96.434
118.06
103.700
122.40
94.945
144.26
2
104.046
93.800
83.636
89.224
116.862
134.30
83.575
91.026
Table (4-6): Main Opinion Score values.
Codecs
types
Network
number
CRLDP
CRLDP
FIFO
CRLDP
PQ
CRLDP
WFQ
RSVP
RSVP
FIFO
RSVP
PQ
RSVP
WFQ
PCM
1
3.5958
3.5958
3.5958
3.5958
3.6019
3.6033
3.6033
3.6033
2
3.6748
3.6748
3.6748
3.6748
3.6745
3.6742
3.6745
3.6745
GSM
FR
1
3.4678
3.4678
3.4678
3.4678
3.4758
3.4758
3.4758
3.4743
2
3.5502
3.5502
3.5502
3.5502
3.5498
3.5498
3.5498
3.5498
G.723.1
1
2.4595
2.4595
2.4595
2.4595
2.4593
2.4593
2.4593
2.4593
2
2.5488
2.5488
2.5488
2.5488
2.5480
2.5480
2.5480
2.5480
G.726
1
3.5961
3.5961
3.5961
3.5961
3.5960
3.5941
3.5960
3.5960
2
3.6752
3.6752
3.6752
3.6752
3.6749
3.6746
3.6749
3.6749
G.728
1
3.5628
3.5628
3.5628
3.5628
3.5959
3.5959
3.5959
3.5959
2
3.6752
3.6752
3.6752
3.6752
3.6749
3.6753
3.6753
3.6755
G.729
1
2.9702
2.9702
2.9702
2.9702
2.9430
2.9430
2.9430
2.9430
2
3.0594
3.0594
3.0594
3.0594
3.0586
3.0590
3.0590
3.0585
IS-641
1
3.4680
3.4680
3.4680
3.4680
3.4758
3.4758
3.4758
3.4743
2
3.5502
3.5502
3.5502
3.5502
3.5495
3.5495
3.5499
3.5499
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
4.4 Summary of Results and Discussion
4.4.1 Number of Maintained Calls
The following points are drawn from Table (4-2)
a. In the case of using FIFO algorithm, the number of maintained
calls decreases when QoS is applied.
b. In the case of using PQ algorithm, the performance of CR-LDP
signal protocol decreases after applying QoS, while the
performance of RSVP signal protocol is enhanced in both
networks.
c. In the case of using WFQ algorithm, the performance of CR-
LDP protocol decreases in the first network and increases in
the second one. However, the significant improvement in the
results is obtained when the RSVP protocol is combined with
WFQ algorithm.
Figures (4-43) and (4-44) show the number of maintained calls in the
first and second networks respectively. Figure (4-45) depicts the number of
maintained calls/sec for each QoS algorithm averaged over the seven audio
codecs types. Figure (4-46) shows the number of maintained calls for each
codec averaged over the four different QoS scenarios.
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Figure (4-43): Number of maintained calls/sec in the first network.
Figure (4-44): Number of maintained calls/sec in the second network.
0
50
100
150
200
250
300
350
CRLDPRSVPCRLDP_FIFORSVP_FIFOCRLDP_PQRSVP_PQCRLDP_WFQRSVP_WFQ
Calls/sec
No. of calls/sec _ net.1
GSM FR G.723.1 G.726 G.728 G.729a IS-641 PCM
0
50
100
150
200
250
300
350
CRLDPRSVPCRLDP_FIFORSVP_FIFOCRLDP_PQRSVP_PQCRLDP_WFQRSVP_WFQ
Calls/sec
No. of calls/sec_net.2
GSM FR G.723.1 G.726 G.728 G.729a IS-641 PCM
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Figure (4-45): Average number of maintained calls/sec for different QoS algorithms for
both networks.
Figure (4-46): Average number of maintained calls/sec as codecs for various codecs.
0
20
40
60
80
100
120
140
160
180
WFQPQFIFOwithout QoS
Avg. calls/sec
Average calls/sec_as QoS comprison
CRLDP_NET.1 RSVP_NET.1 CRLDP_NET.2 RVP_NET.2
0
50
100
150
200
250
GSM FRG.723.1G.726G.728G.729aIS-641PCM
Avg. calls/sec
Average calls/sec_as codecs comprison
Avg. CRLDP_NET.1 Avg. RSVP_NET.1
Avg.CRLDP_NET.2 Avg. RSVP_NET.1
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
4.4.2 Voice Jitter
The following points are drawn from Table (4-3)
a. When FIFO algorithm is applied to the CR-LDP, the results remain
better than that of the RSVP. However, the CR-LDP results without
QoS algorithms are still better than that with FIFO, so applying FIFO
algorithm decreases the performance.
b. The PQ algorithm gives the least jitter values of all other scenarios.
c. In the case of applying WFQ algorithm, the RSVP gives less jitter
than that of the CR-LDP. The results of RSVP are improved, but the
performance of CR-LDP protocol retracted compared to its
performance before applying QoS.
Figures (4-63) and (4-64) show jitter of the first and second networks,
respectively. Figure (4-65) shows the jitter values for each QoS algorithm
averaged over the seven codecs types.
Figure (4-47): Jitter associated with the first network.
-1
0
1
2
3
4
5
CRLDPRSVPCRLDP_FIFORSVP_FIFOCRLDP_PQRSVP_PQCRLDP_WFQRSVP_WFQ
Time (msec)
Jitter NET.1
GSM FR G.723.1 G.726 G.728 G.729a IS-641 PCM
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Figure (4-48): Jitter associated with the second network.
Figure (4-49): Average jitter values (msec).
-0.5
0
0.5
1
1.5
2
2.5
3
CRLDPRSVPCRLDP_FIFORSVP_FIFOCRLDP_PQRSVP_PQCRLDP_WFQRSVP_WFQ
Time (msec)
Jitter NET.2
GSM FR G.723.1 G.726 G.728 G.729a IS-641 PCM
0
0.2
0.4
0.6
0.8
1
1.2
WFQPQFIFOwithout QoS
Time (msec)
Average jitter values
CRLDP_NET.1 RSVP_NET.1 CRLDP_NET.2 RVP_NET.2
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
4.4.3 Packet Delay Variation (PDV)
The following points are drawn from Table (4-4)
a. In the case of applying FIFO algorithm, the performance of PDV is
improved (i.e. its value decreases) in both protocols, and the values
of CR-LDP protocol are still less than that of RSVP protocol.
b. The PQ algorithm gives the least PDV values whether using CR-LDP
or RSVP protocol.
c. In the case of using WFQ algorithm, the PDV decreases whether
using CR-LDP or RSVP signal protocols. However, there is a sudden
jump in the PDV value when using GSM FR codec with CR-LDP
protocol in the first network.
Figure (4-50) and (4-51) show PDV of the first and second networks,
respectively while Figure (4-52) shows the PDV performance for each QoS
algorithm averaged over the seven codecs types.
Figure (4-50): PDV associated with the first network.
0
0.5
1
1.5
2
2.5
3
3.5
4
CRLDPRSVPCRLDP_FIFORSVP_FIFOCRLDP_PQRSVP_PQCRLDP_WFQRSVP_WFQ
Time (msec)
PDV NET.1
GSM FR G.723.1 G.726 G.728 G.729a IS-641 PCM
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Figure (4-51): PDV associated with the second network.
Figure (3-52): Average PDV values (msec) of both networks.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
CRLDPRSVPCRLDP_FIFORSVP_FIFOCRLDP_PQRSVP_PQCRLDP_WFQRSVP_WFQ
Time (msec)
PDV NET.2
GSM FR G.723.1 G.726 G.728 G.729a IS-641 PCM
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
WFQPQFIFOwithout QoS
Time (msec)
Average PDV values
CRLDP_NET.1 RSVP_NET.1 CRLDP_NET.2 RVP_NET.2
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
4.4.4 End-to-end Delay
PQ algorithm improves end-to-end delay in both TE signal protocols.
PCM has the lowest average end-to-end delay and G.723.1 has the largest for
the two networks. Figures (4-53) and (4-54) show the end-to-end delay of the
first and second networks, respectively. Figure (4-55) shows the end-to-end
delay for each codec averaged over the four different QoS scenarios, while
Figure (4-56) shows the end-to-end delay for each QoS algorithm averaged
over the seven codecs types.
Figure (4-53): End-to-end delay values of the first network.
Figure (4-54): End-to-end delay values of the second network.
0
20
40
60
80
100
120
140
160
GSM FRG.723.1G.726G.728G.729aIS-641PCM
Time (msec)
End-to-end delay NET.1
CRLDP RSVP CRLDP_FIFO RSVP_FIFO CRLDP_PQ RSVP_PQ CRLDP_WFQ RSVP_WFQ
0
20
40
60
80
100
120
140
160
GSM FRG.723.1G.726G.728G.729aIS-641PCM
Time (msec)
End-to-end delay NET.2
CRLDP RSVP CRLDP_FIFO RSVP_FIFO CRLDP_PQ RSVP_PQ CRLDP_WFQ RSVP_WFQ
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
Figure (4-55): Average end-to-end delay values for different codecs.
Figure (4-56): Average end-to-end delay values for different QoS.
0
20
40
60
80
100
120
140
GSM FRG.723.1G.726G.728G.729aIS-641PCM
Time (msec)
Average end-to-end delay_as CODECS Comprison
CRLDP_NET.1 RSVP_NET.1 CRLDP_NET.2 RSVP_NET.2
0
20
40
60
80
100
120
WFQPQFIFOwithout QoS
Time (msec)
Average end-to-end delay_as QoS comprison
CRLDP_NET.1 RSVP_NET.1 CRLDP_NET.2 RVP_NET.2
Chapter Four Effect of QoS Algorithms on the Performance of MPLS-TE Networks
4.4.5 Main Opinion Score
The MOS values are clearly affected by audio codec types and the G.723.1
) show the MOS 58-57) and (4-the least values. Figures (4codec type gives
, respectively.networks
nd
2and
st
1for the
Figure (4-57): MOS performance of the first network.
Figure (4-58): MOS performance of the second network
0
1
2
3
4
GSM FRG.723.1G.726G.728G.729aIS-641PCM
MOS Values_net.1
CRLDP RSVP CRLDP_FIFO RSVP_FIFO
CRLDP_PQ RSVP_PQ CRLDP_WFQ RSVP_WFQ
0
1
2
3
4
GSM FRG.723.1G.726G.728G.729aIS-641PCM
MOS Values_net.2
CRLDP RSVP CRLDP_FIFO RSVP_FIFO
CRLDP_PQ RSVP_PQ CRLDP_WFQ RSVP_WFQ
Chapter Five Conclusions and Suggestions for Future Work
Chapter Five
Conclusions and Suggestions for Future Work
5.1 Conclusions
This thesis has addressed the performance of MPLS network in the
presence of TE signal protocols for VoIP applications. Performance
comparison between two TE signal protocols, CR-LDP and RSVP has been
reported. Seven types of audio codecs, namely PCM, GSM FR, G.723.1, G.726,
G.728, G.729, and IS-641, have been used with absence and presence of QoS
algorithms including FIFO, PQ, and WFQ. A number of maintained calls, jitter,
packet delay variation, and end-to-end delay have been used as measure
parameters for performance assessment. The simulation results reveal the
following findings
a. Before using QoS algorithms, the MPLS network with CR-LDP TE
signal protocol has a noticeable performance advantage compared to the
MPLS network with RSVP TE signal protocol. It is five times more than
RSVP in terms of number of maintained calls in the 1st network. This
performance difference increases as the network becomes larger and
becomes seven times in the 2nd network. Further CRLDP has 1.12% less
end-to-end delay than that of RSVP in the 1st network and 8.09% in the
2nd network. For voice jitter, CRLDP is 34.34% of that of RSVP in the
1st network and 8.3% in the 2nd network. The voice packet delay
variation of CRLDP is 21.4% of that of RSVP in the 1st network and
44.88% in the 2nd one.
Chapter Five Conclusions and Suggestions for Future Work
b. After applying QoS algorithms, the WFQ queuing is more compatible
with RSVP than CRLDP protocol. The number of maintained calls in
RSVP-WFQ is (228%) is more than that in CRLDP-WFQ for the 1st
network and (107%) more for the 2nd network. Also, RSVP-WFQ is
more than (CRLDP without QoS) (150%) in the 1st network and (112%)
in the 2nd one. Show Table (5-1).
Table (5-1): The proportion of Number of calls.
2nd Network
1st Network
No. of calls/sec
112 %
150 %
RSVP-WFQ/CR-LDP
107 %
228 %
RSVP-WFQ/CR-LDP-WFQ
c. Using QoS along with TE in MPLS network decreases the jitter, packet
delay variation, and end-to-end packet delay compared to the case of
using TE alone. Further, the priority queuing algorithm gives the lowest
values of the other algorithms.
d. Comparison among codecs shows that G.723.1 gives the highest number
of calls. Thus the highest number of calls is obtained when RSVP, WFQ,
and G.723.1 are used together.
5.2 Suggestions for Future Work
From the work in this thesis, it is found that many ideas can be considered or
investigated in future work. A brief list is given as follows
a. Expanding the work to address the performance of Generalized
Multiprotocol Label Switching (GMPLS) networks incorporating
Chapter Five Conclusions and Suggestions for Future Work
TE signal protocols and QoS algorithms. GMPLS offers
advantages over MPLS such as disassociation of the control and
data channels, scaling the benefits in the control plane, don't
needing routing adjacency between ends of data links, and
supporting many different technologies.
b. Investigating the performance of MPLS network for video
applications in different video formats.
c. Addressing the effect of TE and QoS on the performance of MPLS
Virtual Private Networks (VPNs).
References a
References
[1] M. Bhandure, G. Deshmukh, and J. N. Varshapriya, “Comparative Analysis of Mpls
and Non-Mpls Network”, International Journal of Engineering Research and
Applications, Vol. 3, No. 4, PP. 71-76, Aug. 2013.
[2] A. M. Bongale and N. Nithin, “Analysis of Link Utilization in MPLS Enabled
Network using OPNET IT Guru”, International Journal of Computer Applications,
Vol. 41, No.14, PP. 35-40 , Mar. 2012.
[3] H. G. Perros, “Connection-Oriented Networks SONET/SDH, ATM, MPLS and
Optical Networks”, 1st Ed., John Wiley and Sons Ltd, England, 2005.
[4] M. K. Porwal, A. Yadav, and S. V. Charhate”, Traffic Analysis of MPLS and Non
MPLS Network including MPLS Signaling Protocols and Traffic distribution in
OSPF and MPLS”, Proc. IEEE, PP. 178-192, 2008.
[5] S. Kasera, “ATM Networks Concepts and Protocols”, 2nd Ed. McGraw Hill,
Education, India, 2006.
[6] S. Veni and G. M. Kadhar, “Protection Switching and Rerouting in MPLS”,
International Journal Computer Technology and Applications, Vol. 3, No. 1, PP.
216-220, 2012.
[7] V. Foteinos and K. Tsagkaris, “Operator-Friendly Traffic Engineering in IP/MPLS
Core Networks”, IEEE Transactions on Network and Service Management, Vol. 11,
No. 3, Sep. 2014.
[8] N. H. Almofary, H. S. Moustafa, and F. W. Zaki, “Optimizing QoS for Voice and
Video Using DiffServ-MPLS”, International Journal of Modern Computer Science
and Engineering, Vol. 1, No. 1, PP. 21-32, Nov. 2012.
[9] J. Kharel and D. Adhikari, “Performance Evaluation of Voice Traffic over MPLS
Network with TE and QoS Implementation”, Blekinge Institute of Technology,
Karlskrona, Sweden 2011.
[10] D. Zhang and D. Ionescu,“QoS Performance Analysis in Deployment of DiffServ-
aware MPLS Traffic Engineering,” Eighth ACIS International Conference, on
Software Engineering Artificial Intelligence, Networking, and Parallel/Distributed
Computing, China, PP. 963-967, 2007.
References a
[11] D. Awduche, J. Malcolm, J. Agogbua, M. O'Dell, and J. McManus, “Requirements
for Traffic Engineering over MPLS”, RFC 2702, Sep. 1999. [Online]. Available at:
https://tools.ietf.org/html/rfc2702, last viewer on 26 Mar. 2015.
[12] F. L. Faucheur and W. Lai, “Requirements for Support of Differentiated Services-
aware MPLS Traffic Engineering”, RFC 3564, July, 2003. [Online]. Available at:
https://www.ietf.org/rfc/rfc3564.txt, last viewer on 26 Mar. 2015.
[13] T. Onali., “Quality of Service Technologies for Multimedia Applications in Next
Generation Networks”, PhD Thesis, University of Cagliari, Italy, Feb. 2008.
[14] K. P. Jannu and R. Deekonda, “OPNET Simulation of Voice over MPLS with
Considering Traffic Engineering”, Blekinge Institute and Applications - Workshops,
AINAW 2008. 22nd International Conference, PP. 259-263, 2008.
[15] R. T. Murade, P. M. Ingale, R. U. Kale, and S. S. Sayyad,” Comparative Analysis of
IP, ATM and MPLS with their QoS”, International Journal of Innovative Technology
and Exploring Engineering, Vol. 2, No. 5, Apr. 2013.
[16] H. Hodzic and S. Zoric, “Traffic Engineering with Constraint Based Routing in
MPLS Networks”, Proc. 50th International Symposium ELMAR, Zadar, Vol.1, PP.
269-272, 10-12 Sep. 2008.
[17] T. Aziz, N. Islam khan and A. Popescu ,“Effect of Packet Delay Variation on
Video/Voice over DIFFSERV-MPLS in IPV4/IPV6 Networks”, International
Journal of Distributed and Parallel Systems, Vol. 3, No.1, PP. 27-47, Jan. 2012.
[18] R. S. Naoum, M. Maswady “Performance Evaluation for VOIP over IP and MPLS”,
World of Computer Science and Information Technology Journal, Vol. 2, No. 3, PP.
110-114, 2012.
[19] S. K. Ibrahim and M. M. AL-Quzwini, “Performance Evaluation of MPLS TE Signal
Protocols with Different Audio Codecs for Voice Application”, International Journal
of Computer Applications, Vol. 57, No. 1, PP. 56-60, Nov. 2012.
[20] A. Chaudhary, and S. P. Singh, “ Performance Evaluation of VoIP in MPLS Network
Using NS-2”, International Journal of Computers and Technology, Vol. 13, No. 9,
PP. 4792-4798 , Jun. 2014.
[21] R. Q. Shawl, R. Thaker, E. J. Singh, “A Review: Multi Protocol Label Switching
(Mpls)”, Journal of Engineering Research and Applications, Vol. 4, No.1, PP.66-70,
Jan. 2014.
References a
[22] D. O. Aweduche, “MPLS and Traffic Engineering in IP Networks”, IEEE
Communication Magazine, Dec. 1999.
[23] M. Murakami and Y. Koike, “Highly Reliable and Large-Capacity Packet Transport
Networks: Technologies, Perspectives, and Standardization”, Journal of Lightwave
Technology, Vol. 32, No. 4, PP. 805-816, Feb. 2014.
[24] J. Guichard, and I. Pepelnjak,”MPLS and VPN Architectures”, Cisco Press
Publisher, USA, 2000.
[25] P. Khatri, “MPLS: A Tutorial”, Alcatel-Lucent. Jan. 2014. [Online]. Available at:
www.sanog.org/resources/sanog14/sanog14-paresh-mpls.pdf, last viewer on 26
Mar. 2015.
[26] A. Dumka and H. L. Mandoria, “Dynamic MPLS with Feedback”, International
Journal of Computer Science, Engineering and Applications, Vol. 2, No. 2, PP. 125-
130, Apr. 2012.
[27] S. Kulkarni, R. Sharma, and I. Mishra, “New Bandwidth Guaranteed QoS Routing
Algorithm for MPLS Networks”, Journal of Emerging Trends in Computing and
Information Sciences, Vol.3, No.3, PP. 384-389, Mar. 2012.
[28] T. Chen, T. Oh, "Reliable Services in MPLS", IEEE Communications Magazine,
Vol.37, No.12, PP. 58-62, December, 1999.
[29] P. R. Egli, “MPLS Multi Protocol Label Switching Overview of MPLS, A
Technology That Combines Layer 3 Routing with Layer 2 Switching for Optimized
Network Usage”, [Online]. Available at:
www.indigoo.com/dox/itdp/13_Management&Backbone/MPLS, last viewer on 26
Mar. 2015.
[30] T. Aziz, N. Islam khan and A. Popescu, “Effect of Packet Delay Variation on
Video/Voice over DIFFSERV-MPLS in IPV4/IPV6 Networks”, International
Journal of Distributed and Parallel Systems, Vol. 3, No.1, PP. 27-47, Jan. 2012.
[31] A. Ghanwani, B. Jamoussi, D. Fedyk, P. Ashwood-Smith, Li Li, and N. Feldman,
“Traffic Engineering Standards in IP-Networks using MPLS”, IEEE
Communications Magazine, Vol. 37, No. 12, PP. 49-53, Dec. 1999.
[32] K. Jannu and R. Deekonda, “OPNET Simulation of Voice over MPLS with
Considering Traffic Engineering”, M.Sc. thesis, Electrical Engineering, School of
Engineering, Blekinge Institute of Technology, Sweden, Jun. 2010.
References a
[33] J. F. Ransome, and J. R. Rittinghouse, “VoIP Security”, Elsevier Digital Press, 2005.
[34] J. Martin,” MPLS Based Recovery Mechanisms”, M.Sc. thesis, University of Oslo,
May 2005.
[35] A. Kapoor, S. Godara and S. Khambra, “Comparative Analysis of Signaling
Protocols in MPLS-Traffic Engineering”, National Workshop-Cum-Conference on
Recent Trends in Mathematics and Computing, India, 2011.
[36] Z. Xu,” Designing and Implementing IP/MPLS-Based Ethernet Layer 2 VPN
Services”, Wiley Publishing, Inc., Indianapolis, Indiana, 2010.
[37] A. Kumar and S. G. Thorenoor, “Analysis of IP Network for different Quality of
Service.” International Conference on Computing, Communication, and Control,
Singapore, 2011.
[38] P. Shah, U. Mukhopadhyaya, and A. Sathiamurthi, “Overview of QoS in Packet-
based IP and MPLS Networks”, [Online]. Available at:
https://www.nanog.org/meetings/nanog36/presentations/sathiamurthi.pdf, last
viewer 26 Mar. 2015.
[39] D. D. Clark and W. Fang, “Explicit Allocation of Best-Effort Packet Delivery
Service”, IEEE/ACM Transactions on Networking, Vol. 6, No. 4, PP. 362-373, Aug.
1998.
[40] J. Davidson, J. Peters, M. Bhatia, S. Kalidindi, and S. Mukherjee, “Voice over IP
Fundamentals”, 2nd Ed. Cisco Press, USA, 2006.
[41] K. Wallace, “Voice over IP First-Step”, Cisco Press, USA, 2005.
[42] S. A. Ahson and M. Ilyas, VoIP Handbook: “Applications, Technologies,
Reliability, and Security”, CRC Press, USA, 2009.
[43] B. Forouzan, “Data Communications and Networking”, 4th Ed., McGraw Hill
Higher Education, 2007.
[44] E. Aboelela, “Queuing Disciplines”, Networks Simulation Experiments Manual, 3rd
Ed., USA, 2003.
[45] K. I. Park, “QoS in Packet Networks”, Springer International Series in Engineering
and Computer Science, Vol. 779, 2005.
[46] A. S. Ranjbar. , “CCNP ONT Official Exam Certification Guide, Cisco Press”, May
2007.
References a
[47] W. Odom, J. Geier, and N. Mehta, “CCIE Routing and Switching Official Exam
Certification Guide”, 2nd Ed., Cisco Press, 2006.
[48] M. I. Tariq, M. A. Azad, R. Beuran, and Y. Shinoda, “Performance Analysis of VoIP
Codecs over WiMAX Networks”, International Journal of Computer Science Issues,
Vol. 9, No. 3, PP. 253-259 Nov. 2012
[49] S. Jadhav, H. Zhang, and Z. Huang, “Performance Evaluation of Quality of VoIP in
WiMAX and UMTS”, International Conference on Parallel and Distributed
Computing, Applications and Technologies, Oct. 2011.
[50] K. Salah and A. Alkhoraidly “An OPNET-Based Simulation Approach for
Deploying VoIP”, International Journal of Network Management Int. J. Network,
Vol. 16, No. 3, PP. 159–183, Jan. 2006.
[51] ITU-T recommendation g.107, “The e-model, a Computational Model for Use in
Transmission Planning”, May, 2000.
[52] J. Evans and C. Filsfils,”Deploying IP and MPLS QoS for Multiservice Networks”,
1st Ed., Elsevier Inc, San Francisco, USA, 2007.
[53] B. Goode “Voice over Internet Protocol (VoIP)”, Proc. IEEE, Vol. 90, No. 9, PP.
1495-1517, Sep. 2002.
[54] ”Understanding Delay in Packet Voice Networks”, Cisco Systems, Inc. [Online].
Available at: www.cisco.com/c/en/us/support/.../voice/voice.../5125-delay-
details.html, last viewer on 26 Mar. 2015.
[55] Speech codecs [Online]. Available at:
http://www.couthit.com/embedded_software.asp, last viewer on 22 March, 2015.
[56] ITU-T Recommendation G.723.1, “Dual Rate Speech Coder for Multimedia
communications transmitting at 5.3 and 6.3 Kbit/s”, [Online]. Available at:
www.itu.int/rec/T-REC-G.723.1/e, last viewer on 26 Mar. 2015.
[57] R. Salami, C. Laflamme, J. P. Adoul, and D. Massaloux , “A Toll Quality 8 kb/s
Speech Codec for the Personal Communications System (PCS)”, IEEE Transactions
Vehicular Technology, Vol. 43, No. 3, PP. 808-816, Aug. 1994.
APPENDIX A Voice over Internet Protocol (VoIP)
A-1
APPENDIX A
Voice over Internet Protocol (VoIP)
A.1 VoIP
The generality desired and cost-effective service for all is the implementation of voice
and video (real-time applications) in Internet. VoIP is most commonly transported as
a digitally encoded stream by using Real Time Protocol (RTP) over User Datagram
Protocol (UDP) [52]. RTP is the transport layer protocol contains control part and
data. The control part is a Real Time Control Protocol (RTCP). RTP deals with the
delivery of the VoIP bearer stream from sender to receiver. VoIP packet is transferred
by the collection of IP/UDP/RTP Protocols. Due to the fact that Transmission Control
Protocol (TCP)/IP uses acknowledgement/ retransmission trait, it is not used in real-
time communications, because of it would lead to excrescent delays. In spite of that
it is a reliable communication protocol. TCP/IP are not appropriate for voice
communications, because of that the voice is less permissive to delays, thus UDP is
used with RTP to provide end-to-end transmission of real-time data where RTCP is
employed for monitoring the link [53].
A.2 VoIP codecs
VoIP codecs convert analog voice signals into a digital bit stream at the sender end,
convert them back to analog audio signal at the receiver. Different types of delay
effect of codecs voice as follow [54]
a. Processing (Coder) Delay
The Digital Signal Processor (DSP) takes a period of time for compressing a block of
samples, this delay of time is called coder delay. This is also called Processing del
APPENDIX A Voice over Internet Protocol (VoIP)
A-2
b. Algorithmic delay
The algorithm of compression depends on known voice characteristics to process
sample block N correctly. With a view of reproduce sample block N accurately, the
algorithm should have some knowledge of what is in block N+1. This look ahead,
which is truly an extra delay, and it is called algorithmic delay [54].
c. Packetization delay
The time taken for filling packet payload with encoded/ compressed speech is known
as Pacetization delay. This delay is a result of the sample block size wanted by the
number of blocks placed in single frame voice coder and by the voice coder. Before
the voice samples are released it is accumulate in a buffer, thus the Packetization
delay is also called accumulation delay [54].
For network delay, there are three one-way delay bands considered by the
International Telecommunication Union (ITU) for voice applications, show these
bands in table (A-1).
Table A-1: Delay specifications [54].
Range in
(msec)
Description
0-150
Accepted to user applications mostly.
150-400
Accepted provided that administrators are knowing the
transition time and the effect of it on the transition quality of
the applications of the user.
Over 400
Unaccepted to generic network planning aim. However, in
some especial cases this limit is exceeded.
APPENDIX A Voice over Internet Protocol (VoIP)
A-3
The most widely used codecs are those defined by the ITU G.71x and G.72x
standards. Codecs are more important that determine the quality of voice. There are
important codecs summarizes as follows
A.2.1 G.711 (PCM)
The ITU-T was introduced the G.711 codec in 1972 for digital telephony use.
There is two types of this codec: U-Law is used in Japan and U.S.A. in
international telephone links, and A-Law is used in Europe. G.711 uses a
logarithmic compression. PCM compression ratio is 1:2, because it squeezes
every 16-bit sample to 8 bits. A call expends 128 kbps (with some packet header
overhead) due to that the resulting bitrate for one direction is 64 kbps. PCM
bitrate as a comparison with other codec types is very large. There is no fees
for licensing in this codec, so it can be used in VoIP application freely. The best
work for this codec can be achieved when there is a lot of bandwidth available.
From its advantages that its implementation does not required much power for
CPU (its implementation can be done by using a simple lookup table) and the
quality of audio is a very good quality [55].
A.2.2 G.723
G.723 introduced as an outcome of a competition that announced by ITU to
design a codec allow calls over 28.8 kbps and 33 kbps modem links. The two
different types of G.723 introduced because ITU depended on the two good
solutions and decided to use both of them. Both of them use 30 ms for audio
frames (take 240 samples), however they uses different algorithms. The bitrate
for the first and second type of this codec are 6.4 and 5.3 kbps respectively. The
encoded frames for the first type is 24 bytes and 20 bytes for the second type [56].
APPENDIX A Voice over Internet Protocol (VoIP)
A-4
A.2.3 G.726
In 1990 the G.726 codec type was introduced by ITU-T. The codec converts a
G.711 encoded speech signal sampled at 8 KHz.to a.16, 24, 32, or 40 kbps bit-
stream respectively and vice-versa. G.726 depend on Adaptive Differential
Pulse Code Modulation (ADPCM) algorithm [55]. This codec is the standard
codec used in Digital Enhanced Cordless Telecommunications DECT) wireless
phones.
A.2.4 G.728
In 1992 ITU-T was standardized a G.728 audio codec. This codec depends on
Low-Delay Code-Excited Linear Prediction Coding (LD-CELP) algorithm.
Operates on 16 bit speech signal and uses 8 KHz to sample the signal and 16
kbps bitrate used to generate compressed bit stream. An inherent packet loss
concealment mechanism used in the decoder of this codec type. This codec
used in satellite telephony, voice over cable applications, and video
conferencing. [55].
A.2.5 G.729
G.729 codecs provides a good audio quality although that it's require a low
bandwidth. The frame uses in this codec is 10 msec long for encoded audio.
The 10 msec frame includes 80 audio samples, given the sampling frequency
of 8 kHz. The bitrate of G.729 is 8 kbps for one direction, because of each
frame encode to 10 bytes in this codec algorithm. This codec is licensed. G.729
codec uses the Conjugate Structure – Algebraic Code Excited Linear Prediction
(CS-ACELP) algorithm [55].
APPENDIX A Voice over Internet Protocol (VoIP)
A-5
A.2.6 GSM FR
GSM-Full Rate (GSM-FR) speech codec was introduced in early 1990s and
adopted by the 3rd Generation Partnership Project (3GPP). This codec type for
mobile telephony. In this codec the speech frame is 20 ms, each frame sampled
at 8 KHz and the compressed bit streams are generated with 13 kbps average
bitrate. To compress speech this codec uses the Linear Predictive Coding with
Regular Pulse Excitation (LPC-RPE codec) algorithm. This codec provides
Voice Activity Detection (VAD) and Comfort Noise Generation (CNG)
algorithms, and an inherent Packet Loss Concealment (PLC) algorithm for
handling frame erasures. This codec was particularly advanced for mobile
telephony through GSM network [55].
A.2.7 IS-641
IS-641 speech coding is based on the ACELP (Algebraic-code-excited linear
prediction). The bit rate of the speech codec is 7.4 kbps. The standard has been
superseded by the Telecommunications Industry Association (TIA) / Electronic
Industries Alliance (EIA) -136-410 [57].
ةــــــــــــــــــــــــــــــــــــــــــــصخلا
(MPLS)
(TE)
VoIP
CR-LDP RSVP
Switch
PCM (64 Kbps) GSM FR
(13Kbps) G.723.1 (5.3 Kbps) G.726 (16 Kbps) G.728 (16 Kbps)G.729 (8
Kbps) .IS-641(7.4 Kbps)
(QoS)FIFO PQWFQ
JitterPDVMOS
CR-LDP
RSVP
CR-LDPRSVP
WFQ
RSVPCR-LDPWFQ
RSVPCR-LDP
jitter
PQ
G.723.1RSVP
WFQOPNET
2011
1436
2015
