International Journal of Pure and Applied Mathematics

Volume 119 No. 10 2018, 1171-1186

ISSN: 1311-8080 (printed version); ISSN: 1314-3395 (on-line version)
url: http://www.ijpam.eu
Special Issue
ijpam.eu

Study of QoS efficiency in Deployment of

MPLS and MPLS/ DiffServ
Faycal
1
Bensalah1*, Najib EL Kamoun1,Ayoub Bahnasse2
Lab STIC, Faculty of Sciences El Jadida , University Chouaib Doukkali ,Morocco ,
f.bensalah@ucd.ac.ma , elkamoun.n@ucd.ac.ma ,
2
Laboratory LTI, Dept.Physics, Faculty of Sciences Ben M’sik, University Hassan II, Morocco
a.bahnasse@gmail.com

Abstract
The Multi-Protocol Label Switching (MPLS) protocol has contributed to the
Internet routing, traffic engineering and quality of service required for new services.
It would be interesting to compare the QoS performance of MPLS and MPLS /
DiffServ networks, taking into account their particular constraints.
In this article, we evaluated the QoS performance metrics such as delay variation,
delay, response time, throughput for different traffic types (voice, data and video) for
both platforms MPLS and MPLS / DiffServ.
The objective is to compare the performance of MPLS and MPLS / DiffServ using
"OPNET Modeler v14.5" using the latest simulation techniques, where different QoS
parameters can be measured to compare the performance of networks.
Our approach in this work is to design and build an operator network typebackbone
to simulate a real scenario that conveys different types of traffic (voice, data and
video).
The results of the work are presented according to the simulation time and the
network load. The results of the comparison demonstrate the advantage over the
performance of MPLS networks with diffserv compared to traditional MPLS networks.

Keywords:, MPLS, DiffServ, Qos, NGN

1. Introduction
The objective of our research is to present first the characteristics of the architecture
of NGN networks, and to make a detailed study for the implementation of a core
network based on an IP / MPLS platform.

1.1. NGN


NGN or Next Generation Network in English (literally "New Generation
Network") is a phrase frequently used in the telecommunications industry,
especially since the early 1990s[1].

the term most often refers to the network of a telecommunications company

whose architecture is based on a packet transfer plan capable of replacing the
switched telephone network and other traditional networks. The operator has a

1171
International Journal of Pure and Applied Mathematics Special Issue

unique network core that enables it to provide subscribers with multiple services
(voice, data, audiovisual content, etc.)

NGN networks rely on an architecture in independent layers (transport, control,

services) communicating via open and standardized interfaces. Services must be
scalable and accessible independently of the access network used[2].

1.2. Convergence to MPLS

Multiprotocol Label Switching (MPLS) is often considered one of the major
technologies to implement quality of service in packet switched networks.
However, MPLS was originally developed by the IETF, with the objective of
establishing a common standard for transporting IP packets on switched s ubnets.
MPLS networks are "connection-oriented" networks enabling traffic
engineering (Traffic Engineering, MPLS-TE) of packet-switched networks. As
such, they can guarantee bandwidth for various streams, which is the first
condition for providing collateral.
The goal of MPLS is to give IP routers greater switching power by basing the
routing decision on label information (or tag) inserted between Data Link Level
2 and Network Layer 3 ) [3].
Today, the real motives for deploying MPLS solutions are the applications that
MPLS allows, and that were very difficult or even impossible to implement with
traditional IP. These applications are very important for operators and ISPs
(Internet Service Providers), simply because they can be sold.
There are now four major applications of MPLS. These applications involve
the implementation of components adapted to the desired functionalities. The
implementation of MPLS will therefore be different depending on the objectives
sought. This is mainly reflected in a different way of assigning and distributing
labels (Classification, Labels Distribution Protocols). The principle of routing
packets based on the use of labels being the basic mechanism common to all
approaches .
The main applications of MPLS are:

 Any Transport over MPLS (AToM).

 Support for virtual private networks (MPLS VPN).

 Support for quality of service (MPLS QoS).

 Traffic Engineering (MPLS TE).

MPLS allows to efficiently solve the tunneling functionality, since the routing
of the packets is not carried out on the destination address of the IP packet but
on the value of the label assigned to the packet.[4].

1.3. Quality of service support (MPLS QoS)

1172
International Journal of Pure and Applied Mathematics Special Issue

With MPLS technology, QoS [5-6] is a crucial element for an operator network.
Indeed, the operator must be able to guarantee to his customers the transport
of their flows by guaranteeing different constraints. The quality of service is
mainly divided into four parameters: throughput, delay, jitter and loss.

- The throughput represents the transmission resources occupied by a stream. A

stream is a set of packets resulting from a user application.
- The delay is the end-to-end transfer time of a packet.
- The jitter corresponds to the latency variations of the packets. The jitter
originates essentially from traffic variations on the output links of the routers.
- Packet losses may be due to data integrity errors or packet rejections in the
event of congestion.
Support for QoS can be implemented in two ways on MPLS:
o Diffserv (Differentiated Services): Traffics on the same LSP can be
assigned to different queues in the LSR routers, depending on the value of
the EXP field of the MPLS header.

o The use of Traffic Engineering.

1.4. Differentiation of services (MPLS Diffserv)

The service differentiation model [7] appears to be more appropriate for

multiservice networks such as the Internet. In other words, this model means
giving priority to a class of service at the expense of another class at the time of
congestion. The DiffServ (Differentiated Services) model defines a completely
different approach compared to the IntServ (Integrated Services) model. It does
not require an end-to-end reservation or signaling. It allows each packet to be
assigned to a class of service. Complexity is relegated to the ends of the n etwork.
The differentiated services of the DiffServ architecture make it possible to
substantially reduce the state information that each node of the network must
memorize.
In the DiffServ architecture, the differentiated processing of the packets is based
on 3 fundamental operations:

• Classification of flows into classes of services.
• The introduction of priorities within classes (Scheduling).
• Traffic management in a given class (Queue management).

Diffserv defines four PHB [8] or service classes:

 Best Effort (priorité basse) : 

 Expedited Forwarding (EF) (RFC 2598)
 Assured Forwarding (AF) (RFC 2597)

1173
International Journal of Pure and Applied Mathematics Special Issue

 Default Forwarding (DF)

This notion of PHB makes it possible to construct a variety of differentiated
services. PHBs are implemented by constructors in routers using queuing
mechanisms (Custom Queuing, Weighted Fair Queuing, ...) and flow control.


1.5. Traffic Engineering (MPLS TE)

This application is closely related to the quality of service, since its immediate result
is the improvement of parameters such as delay or jitter in the network. It is still
regarded as an application in its own right by most manufacturers. This is because
MPLS TE[9] is not a simple resource reservation technique for network applications.
It is a more general concept that aims to be a solution that aims to increase the overall
performance of the network by playing on the balanced distribution of loads (traffics)
in the network so as to have a more optimal use of the links.

2. Scenario of the experiment

In this modeling we used OPNET Modeler 14.5 to compare the performance of
MPLS_Diffserv technology against traditional MPLS, two scenarios of network cores,
configuring (03) application types (Ftp, voice, video conferencing).

2.1. Simulation Parameters

We have implemented a backbone architecture based on 9 network router (P) attached

with 34MB E3 links and two Edge (PE) routers and attached to the core router with
2M E1 links, architecture and deploy with the MPLS_Diffserv and MPLS in order to
make a comparative study between the 2 scenarios and the quality of service provided
by each technology.

Figure 1 below shows the physical topology of the network of this simulation

1174
International Journal of Pure and Applied Mathematics Special Issue

Figure 1 : Scenario1_Backbone MPLSFigure 2 : Scenario2_Backbone

MPLS_diffserv
Figure 2 below shows the MPLS_diffserv network, which is the same as the MPLS
network for this search by adding the QOS object.

2.2. Configuring an MPLS Network

a) Source of Trafic :

The traffic sources are generated by the client nodes of the objects (node) from which
we associate the application profile that presents the configuration of multiple
applications. We used Ftp, voice and video in our simulation.

The figures below show the chosen values and parameters of our traffic model.

Trafic Ftp :For Data traffic (FTP), the service type is AF11
Trafic Voice :For voice traffic, the encoder is GSM FR, the service type is AF31
Trafic Video:For video traffic, we use an average resolution of 15 frames / sec
(frame / sec) of 128x240 pixels, setting the DSCP field to AF41.

b) DSCP to EXP Mapping :

The network we have set up allows us to achieve a continuity of service between the
IP networks (representing the local networks) and the MPLS network (representing
the Operator network).To ensure this continuity, we use the DSCP to EXP mapping
table[10]

This principle consists in copying the value of the ToS field of IP in the EXP label of
MPLS. It is used by the LER routers (as input to the MPLS network). It ensures the
"translation" of the class values between IP and MPLS.

1175
International Journal of Pure and Applied Mathematics Special Issue

Once the MPLS network is schematized, it must be configured so that it is operational

according to selected criteria. The steps to configure an MPLS network are:
- the definition of FEC.
- the definition of the Trunks
- the definition of LSPs
- configuration of MPLS switches

c) The value of the ToS field

The ToS field [11] of an IP packet consists of 8 bits. The value of this field
can be configured according to the approach, as defined in the original IP
specification using four parameters [RFC 791].
These parameters are Delay, Throughput, Reliability and Precedence. The
Precedence parameter defining the importance of the datagram can take the following
8 values: (0) Best Effort, (1) Background, (2) Standard, (3) Excellent Effort, (4)
Streaming Multimedia, Multimedia, (6) Interactive Voice and (7) Reserved.

The ToS field of IP packets can also be configured using the Differenciated Services
Code Point (DSCP) approach [RFC 2474] used with IP networks supporting the
Diffserv quality of service architecture. In this case, the ToS field takes one of the
values defined for the different service classes of this architecture.

The value of this field can then be: Expedited Forwarding (EF), Assured Forwarding
(AF1, AF2, AF3, AF4). If the service class of the packet is AF1 to AF4, there is
added a value defining the priority to be deleted if necessary: AFx1 for packets
having priority not to be deleted, AFx2 for packets which can be deleted at need,
AFx3 for packets of class x to be deleted first.

d) Definition of Traffic Trunk

A Traffic trunk [12] is not part of the foundation of MPLS technology. In an MPLS
network without traffic engineering, the packets characterized by a FEC follow the
corresponding LSP. Traffic Trunk is a concept related to traffic engineering. To move
traffic where there is bandwidth, the FEC associates traffic to another LSP, it is
Traffic Trunk that is associated with another LSP. When traffic engineering is used,
FEC associates traffic to a Trunk Traffic that is associated with one or more LSPs.
OPNET requires the use of Traffic Trunk. To configure those used in our simulation,

1176
International Journal of Pure and Applied Mathematics Special Issue

we must edit the Traffic Trunk Profiles attribute of the MPLS Configuration object.

For our simulation, we have configured a Trunk profile for voice traffic and another
for video and other for FTP. The Trunk profile associated with the voice is
characterized by any maximum and average bit rate. Indeed, its capacity will have to
vary according to the needs determined by the prediction of the bandwidth
requirements determined by the admission control function.

The excess traffic will not be deleted and the Trunk's service class is Expedited
Forwarding to ensure the desired bandwidth and minimize the delay and jitter of voice
packets.

For the video, a different Trunk will be used to mark this traffic with a different class
of service. This Trunk will have a maximum and average throughput of 64,000 bps.
Excess traffic will not be removed and the class of service will be Assured
Forwarding 41.

e) Association of FECs in LSP

Each traffic type must be associated with the corresponding specific LSP [13] , and
which carries traffic to the destination PE router. The association of traffic to the LSP
is performed on the LER_1 (Traffic Mapping Configuration) device router (see figure
below)

Figure 3 :Association of FEC in LSP

Association of FECs to LSP for FTP Service

FTP traffic will be routed through the LSP (in Green) (LER1 LSR_1
LSR_4LSR_7LSR_8LSR_9LER2)

1177
International Journal of Pure and Applied Mathematics Special Issue

Association of FECs in LSP for voice service

Voice traffic will be routed through the LSP (in Red) (LER1 LSR_1
LSR_2LSR_3LSR_6LSR_9LER2)

The voice service does not have a backup LSP (in case of a break)

Association of FECs at LSP for video service

The video traffic will be routed through the LSP (in Blue) (LER1 LSR_1
LSR_5LSR_9LER2)

f) Configuring MPLS Switches

The activation of MPLS differs depending on the location of the router in the
backbone, in the 9 P routers we have enabled MPLS on all interfaces while in both PE
routers MPLS is enabled only on interfaces linking these routers to routers P We
chose the LDP protocol to distribute the MPLS labels.

3. Analysis of simulation results

3.1. Traffic Analysis Data (FTP)

Figure 4 : FTP response time with light load (Blue - MPLS, Red - MPLS /
DiffServ)

The MPLS response time is higher than the MPLS / DiffServ network. We notice that

1178
International Journal of Pure and Applied Mathematics Special Issue

the MPLS network with DiffServ provides better performance in response times for
FTP-based traffic.

Figure 5 : FTP traffic response time with heavy Load (Blue - MPLS / DiffServ; Red-
MPLS)

After increasing the load on the network (heavy load), the response time for MPLS
changes to a very high value, but DiffServ remains about the same as shown in Figure
5.

3.2. Traffic Analysis Voice:

Then we pass on the study of network performance when traffic is the voice. We will
compare between the two scenarios MPLS and MPLS_ DiffServ. Figure 6 and Figure
7 illustrate the end-to-end delay parameters for light load and heavy load traffic.

Figure 6 : End-to-end traffic delay Voice with light load (Blue - MPLS; Red - MPLS /
DiffServ)

1179
International Journal of Pure and Applied Mathematics Special Issue

Figure 7 : Delayed end-to-end traffic Voice with heavy load (Blue - MPLS / DiffServ;
Red-MPLS)

Figures 8 and 9 show the delay (jitter) variations of voice traffic for light load and
heavy load traffic.

Figure 8: Variation of delay time (jitter) of traffic Voice with light load (Blue - MPLS;
Red - MPLS / DiffServ)

10

1180
International Journal of Pure and Applied Mathematics Special Issue

Figure 9: Variation of the Delay Delay (Jitter) of Voice heavy load traffic (Blue -
MPLS, Red-MPLS / DiffServ)

3.3. Analyse du trafic Video :

Then we study the performance of the network when the traffic is the Video. We will
compare between the two scenarios MPLS and MPLS_ DiffServ. Below, illustrate the
end-to-end delay settings for light load and heavy load traffic.

Figure 10 : Video End-to-End Delay with Light Load (Blue - MPLS; Red - MPLS /
DiffServ)

11

1181
International Journal of Pure and Applied Mathematics Special Issue

Figure 11 : Video End-to-End Delay with heavy Load (Blue - MPLS / DiffServ; Red-
MPLS)

The end-to-end delay for video stream is higher for MPLS than MPLS / DiffServ.
After the load increases, the end-to-end delay for MPLS becomes very high and keeps
increasing. While the MPLS / DiffServ delay remains at a low level. This shows that
MPLS / DiffServ offers better quality even in congested network with higher loads,
while in MPLS, the delay increases to a very high value.

Figure 12 : Variation of the delay time (jitter) of video traffic with light load (Blue -
MPLS; Red - MPLS / DiffServ)

The variation of the delay in FIG. 13 demonstrates the best quality of the MPLS /
DiffServ service.

12

1182
International Journal of Pure and Applied Mathematics Special Issue

Figure 13 : Variation of delay of video traffic with heavy Load (Blue - MPLS /
DiffServ; Red - MPLS)

After the increase in traffic load, the results of FIG 13 are evident that the delay
variation has increased to a very high value for MPLS, while the variation of the
MPLS_ DiffServ delay has remained at a very low value, keeping the quality of
service performance to the required level.

3.4. The parameters were compared

• FTP traffic response time (light / Heavy Load)

• End-to-End Video Traffic Delay (light / Heavy Load)
• Variation of delay with video traffic (light / Heavy Load)
• End-to-End Voice Traffic Delay (light / Heavy Load)
• Variation of video traffic delay (light / Heavy Load)

For all the parameters discussed above, MPLS with DiffServ demonstrates better
performance on both Heavy load and light compared to MPLS.

The response time of FTP traffic on MPLS_ DiffServ with Heavy Load was lower
than MPLS.

The delay of voice and video traffic as shown in Figures 6-13 demonstrates that
MPLS / DiffServ has a lower delay value than MPLS; the value of the delay on the
Heavy Load model becomes much more important.

13

1183
International Journal of Pure and Applied Mathematics Special Issue

The delay values for voice and video are less than 150 ms, according to the ITU-T
standard.

4. Conclusion
Consider in our simulations that the combination of DiffServ and MPLS presents a
very attractive strategy for the network service providers since it makes it possible to
ensure the load sharing and the quality of service demanded by the customers.
However, managing this type of network is not a simple function and can not be done
manually.
and it is in this sense that our perspective presents a new archetecture of automation
of MPLS management with diffServ based on the SDN.

5. References
5.1. Journal Article
1. Ito, A., & Fukuda, K. (2016). Next-generation Network Architecture Led by Information-
Centric Networking. FUJITSU Sci. Tech. J, 52(1), 53-60.

2. FAROOQ, Muhammad, AHMAD, Abid, et ULLAH, Irshad. QoS Mechanisms in

NGN. International Journal of Engineering Works, 2017, vol. 4, no 5, p. 2409-2770.

3. BENSALAH, Faycal, EL KAMOUN, Najib, et BAHNASSE, Ayoub. Scalability Evaluation of

VOIP over Various MPLS Tunneling under OPNET Modeler. Indian Journal of Science and
Technology, 2017, vol. 10, no 29.

4. BENSALAH, Faycal, EL KAMOUN, Najib, et BAHNASSE, Ayoub. Evaluation of tunnel layer

impact on VOIP performances (IP-MPLS-MPLS VPN-MPLS VPN IPsec). International
Journal of Computer Science and Network Security (IJCSNS), 2017, vol. 17, no 3, p. 87.

5. ADEWALE, Adeyinka A., MATTHEWS, Victor O., AGBOJE, O. E., et al. Performance
Enhancement of Quality of Service of IP Mobility for Real-Time Traffic. International Journal
of Computer Science and Telecommunications, 2017, vol. 8, no 3, p. 13-17.

11. DUMKA, Ankur et MANDORIA, Hardwari Lal. Enhancement to performance of MPLS

Network through hierarchical MPLS. International Journal of Communication Networks and
Distributed Systems, 2017, vol. 19, no 1, p. 19-27.

5.2. Book

9. ROBERTAZZI, Thomas G. Multiprotocol Label Switching (MPLS). In : Introduction to

Computer Networking. Springer International Publishing, 2017. p. 61-65.

10. GEIB, Ruediger et BLACK, David. Diffserv-Interconnection Classes and Practice.

2017.

12. AWDUCHE, Daniel O. et AGOGBUA, Johnson. Requirements for traffic engineering

over MPLS. 1999.

14

1184
International Journal of Pure and Applied Mathematics Special Issue

13. BERZIN, Oleg et MALIS, Andrew. Mobility support using MPLS and MP-BGP
signaling. 2008.

5.3. Chapter in a Book

6. MAHMOUD, Nura AbdAlrhman Alhaj Alsharief, et al. QOS Performance Analysis in

deployment Differentiated Service with Multiprotocol Label Switching for Voice over
IP. 2017. Thèse de doctorat. Sudan University of Science and Technology.
7. PERROS, Harry G. QoS Architectures for the IP Network. In : Encyclopedia of
Information Science and Technology, Fourth Edition. IGI Global, 2018. p. 6609-6617.

5.4. Conference Proceedings

5. TAHIR, Junaid, SIDDIQI, Muhammad Zain, et ARIF, Sharif. Performance analysis of

MPLS based networks with conventional networks. In : Recent Trends in
Telecommunications Research (RTTR), Workshop on. IEEE, 2017. p. 1-4.

Authors

Faycal Bensalah received the Master degrees, Network and

telecommunication, from Faculty of sciences El Jadida in 2014.
Network administrator at Chouaib Doukkali
University,Actually a Ph.D Student on STIC Laboratory on
Faculty Of sciences El Jadida, Network and
Telecommunications team. His research interest are : NGN,
MPLS , Networks, QoS on mobile networks, wireless
networks, networks and telecommunications.

Najib ElkamounPh.D, professor higher education degree at

Faculty of sciences El Jadida.in the dept. of physics.
Researcher member on STIC laboratory, header of Network
and Telecommunications team. His research interest includes,
NGN, MPLS , Networks, QoS on mobile networks, wireless
networks, networks and telecommunications.

Ayoub BAHNASSESince joining the Software Engineering and

Telecommunications Team, on LTI Laboratory, Faculty of
Sciences Ben M’SIK, University Hassan II Casablanca,
Ayoub BAHNASSE has been involved with studies
related New Generation Networks, Networks security and
Mobile Learning. Before joining University, BAHNASSE
obtained Ph.D degree on University Chouaïb DOUKKALI El
Jadida on 2016. Actually BAHNASSE was awarded as
an outstanding reviewer on Elsevier Computer Network
journal and technical program committee on several
international conferences: Recent Trends in Computer Science
and Electronics, EAI International Conference on Technology,
Innovation, Entrepreneurship and Education. My research
fields are not limited only on: Security, mlearning, Wireless
networks, QoS , IMS and NGN, ioT, smart city, MPLS

15

1185
1186