Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/312020851

Performance Analysis of Voice Over LTE Using OMNeT++

Article · August 2016

CITATIONS READS

3 2,780

2 authors:

Maryam Shafei Ahmed J Jameel

2 PUBLICATIONS   12 CITATIONS   
Ahlia University
77 PUBLICATIONS   181 CITATIONS   
SEE PROFILE
SEE PROFILE

Some of the authors of this publication are also working on these related projects:

A Multicore Reconfigurable Processor Architecture for Energy and Throughput Aware Applications View project

Single Sideband (SSB) View project

All content following this page was uploaded by Ahmed J Jameel on 01 January 2017.

The user has requested enhancement of the downloaded file.


Performance Analysis of Voice Over LTE Using
OMNeT++

Maryam M. Shafiei Ahmed J. Jameel


Department of Information Technology Department of Telecommunication Engineering
Ahlia University Ahlia University
Manama, Kingdom of Bahrain Manama, Kingdom of Bahrain
m.shaffieei@hotmail.com adulaimi@ahlia.edu.bh

Abstract—This paper describes the performance analysis of implemented so the voice calls will automatic shift and full
voice over LTE using OMNeT++; an open-source system-level back to the best old bearer available (2G and 3G) [2].
simulator for support LTE networks. By using SimuLTE is based
on OMNeT++, a well-known, widely-used modular simulation
framework, which offers a high degree of experiment support. As
II. LTE NETWORK ARCHITECTURE
such, it can be seamlessly integrated with all the network The high level of the LTE network architecture as shown in
oriented modules of the OMNeT++ family, such as INET, thus Figure 1 is mainly composed of three main components:
enabling – among other things. We describe the voice over LTE,
and show performance evaluation results obtained using the
simulator.

Keywords— OMNeT++; INET; VoLTE; LTE; Voice Over LTE


Simulation

I. INTRODUCTION
In the past few decades, the mobile telecommunication
industries have worked very hard to shift between each Fig. 1. LTE Network Architecture
generation ranging from the 1G to LTE. This conversion from
1G to 4G was not as easy as it took a lot of work to make 4G A. UE (The User Equipment)
technology the fastest network rollout. Long Term Evolution
(LTE) was designed for the data transfer and also as a packet This component mainly consists of few of the
switched all-IP system. It does not contain any circuit switched functionalities of Mobile Terminal (MT) that is held
domain for the purpose of providing with the regular voice and responsible for all kinds of functioning of the call. On the other
SMS services. The increase of the data traffic raised the issue hand, the Terminal Equipment, which is also considered as one
of mobile broadband services by the consumers. The latest of of the major devices of UE serves the function of data
these developments is the voice over LTE (VoLTE) it is streaming and Universal Subscriber Identity Module (USIM).
devised scheme for standardized system between the mobile The USIM stores the network identification and user
operators to carry out voice over Long Term Evolution (LTE) information. In this simulation of the LTE network, the User
technology by replacing voice over the old technologies. The Equipment such as mobile, tablet, laptop …etc. has been used
idea of the voice over LTE based on simply adapt to a [1].
completely new infrastructure based on internet protocol (IP) to
replace the old legacy (2G-3G). The VoLTE specifications are B. E-UTRAN
based on air-interface, which is based on orthogonal frequency The Evolved UTMS Terrestrial Radio Access Network
division multiplexing (OFDM) [1]. used to handle the radio communication between the user
The new technology VoLTE can provide a combined equipment (UE) and the EPC. E-UTRAN composed of one or
system for transfer the voice traffic over the long term more base station called eNB or eNodeB. One of the parts of
evolution air network access and employ the voice-over-IP E-UTRAN is also termed as The Radio Access Network
(voice over internet protocol) technology, which is based on (RAN). The eNB or eNodeB serves the function of providing
the (IP)- multimedia and IMS sub system to provide an E-UTRA user plane and also it controls the plane protocol
appropriate service and video calling. The setup protocol for terminators along the user equipment [3].
connection control is the session initiated protocol (SIP) which
built to work with generic open IP network. The LTE is C. EPC (Evolved Packet Core)
acclimatized with the current networks (3GPP, GSM,
WCDMA, HSPA) and support for full forward and backward Evolved Packet Core is composed of two main elements:
compatibility, until the LTE network voice service is fully The Service Gateway (S-GW) which allows the user to
communicate with other users of LTE network and PDN OMNeT++ as the major tool for the simulation is that it is one
Gateway (P-GW) which is responsible to provide the of the most mature, stable and enriched with features
connectivity between UE and external network like Internet. It framework [4
serves the function of controlling the network access,
management of mobility, and the other functions of network B. INET Framework
management. The Home Subscriber Server (HSS) present in OMNET++ has some of external extensions that can be
the EPC stores all the information related with the subscriber. used to design and simulate the wireless network such as INET
The entity of management of mobility controls the release and Framework. The INET framework is an open-source model
set-up of connections existing between the packet data network that should be installed on top of OMNET++. In addition to the
and user. It also accomplishes its activity through the wireless network, it can be used to simulate wired and mobile
registration of UE authentication location and using valuable networks. It contains IPv4, IPv6, TCP, SCTP, UDP protocol
information from the HSS. The Packet Data Gateway (P-GW) implementations and some of the other application models. As
does the function of GGSN and SGSN, which also signifies the that of OMNeT++, INET framework also uses the similar
connectivity to the IP network. This system is assigned with modules that communicate through the passing of message. [5]
the varied tasks of assignment of IP address, DHCP functions,
user authentication, Quality of Service (QoS), charging data
creation Deep Packet Inspection (DPI) [3]. C. SimuLTE
SimuLTE for OMNeT++ can be used to analyze and
evaluate the performance of LTE and LTE Advanced
III. SIMULATOR OVERVIEW
networks. It is an open source project developed by group of
Simulation software accomplishes a major role in the researchers to evaluate the complex network environments. It
analysis of complex automation system and non-linear control should be installed on top of OMNET++ and INET
system. Few of the software of computer that are designed for Framework. It simulates the data plane of the LTE Radio
the dynamic system simulation at higher level than that of Access Network and Evolved Packet Core. SimuLTE has a
programming languages can be named as simulation languages, special feature that it contains around 40000 lines of codes that
simulation software, simulation system, simulation helps in the extra functionalities such as the applications,
environment and the simulators. Basically, simulation is mobility, event queues, ID/UDP, and so on. However, in this
explained as a particular method which is used for the solving particular study of the voice over LTE, this distinguished
of a problem in the dynamical systems, and which also finds feature of SimuLTE has been extracted from the OMNeT++
out the model of the system rather than the real system. and INET frameworks. [5].
Simulation process follows few of the steps in sequence, which
can be listed as formulation of problem, collection of data,
IV. SIMULATION
mathematical modelling, identification of the model,
experiments with the model, representation of the result and This section describes the implemented simulation
interpretation of the result. Simulation software is usually used topology in OMNET++, GUI and explains the simulation
for designing, studying and analyzing the network parameters used in the experiment.
communications. There are various software’s available in the
market that can serve this purpose. Most of the simulation A. Quality of Service (QoS) Criteria
software’s are commercial but some of them are free for non- The performance of Voice over LTE can be measured with
commercial use such as OMNET++. the help of various criteria. In this experiment the major focus
will lie on the following four of the major criteria.
A. OMNeT Framework
1) Mean Opinion Score (MOS):
OMNET++ is and extensible open-source library and
MOS is the grading system that is used for the measuring of
framework primarily used to simulate networks. It can be used the quality of a voice call. It is usually graded by the user with
in various problem domains such as modelling of wired and the scale of 1 to 5, which means bad to excellent. This
wireless communications networks, evaluating performance particular score is determined by few of the factors such as end
aspects of complex networks. OMNET++ is widely used by to end delay, jitter and packet loss. One of the empirical
academic institutions and educational environments for formula that can be used for the calculation of MOS score from
teaching purpose. It’s also used by students and researchers to the packet loss in terms of percentage in milliseconds is as
study and analyze the performance of the networks. follows [6].
The basic building block of OMNeT++ is modules, either
simple module or compound module. These modules MOS  ln( loss )  0.1 ln( size ) (1)
communicate through messages that are sent and received
The following table shows the standard and the ideal
through connection linking the gates of the modules.
quality values for the Mean Opinion Score (MOS).
OMNeT++ facilitates the user to keep the implementation,
description and parameter values of the model separate. C++ is
used as the coding of the implementation. The files written in TABLE I. MOS STANDARD
Network Description (NED) language is used for expressing
MOS Quality
the description. Theses NEDs also allow for writing of the
parametric topologies. The major reasons for selecting 5 Excellent
MOS Quality solved with the addition of jitters buffers. This is an important
4 Good parameter to be considered while measuring the quality of
service since the high jitter can lead to poor quality of voice.
3 Fair
The high jitter usually leads to the weaker quality of call as the
2 Poor information of the voice will not be received within the timely
manner and thus, the information will not make any sense. In
1 Bad
the technical terms, jitter is the measure of the variability of the
latency over the time and also across the network [7].
2) End to End Delay : The jitter that exists between the starting and final point of
End to End Delay is the time taken for a voice packet to be the communication must always be less than 100ms. If the
transmitted from the source UE to the destination UE across value of the jitter becomes smaller than 100ms, it can be
the LTE network. In simple words, it can be explained as the adjusted with the addition of jitter buffers [7]. The following
difference in the time between the sending and receiving of the table shows the average and the ideal quality values for the
packet. It basically takes place due to the performance of the Jitter:
network and the distance that exists between two of the nodes.
This parameter is crucial so as to receive more information on TABLE IV. JITTER STANDARD
the voice of a real time. There would be difficulty in having the
effective communication in case of too much delay. Jitter Quality

< 20 ms Ideal
The following table shows the average and the ideal quality
values for the VoLTE End to End Delay. < 50 ms Average

TABLE II. END TO END DELAY STANDARD

End to End Delay Quality B. Simulation Configuration


< 50 ms Ideal
This section presents all the general parameters used in the
conduction of the simulation.
< 150 ms Average
 Ethernet Link Data Rate: 10Mbps
 Simulation Time: 20 seconds
3) Packet Loss :
Packet Loss can be defined as the number of the transmitted  Packet Size: 40 byte
packets that are failed to reach its destination. It can also be  Queue Size: 1MB
described as the particular rate in which the packets that are
being sent do not reach at the receiving end. The real time C. Voice Over LTE Scenarios
communications are based on the UD protocols. This protocol
is usually without any connections and it cannot be send again 1) Scenario (1 & 2) Voice Over LTE network.
OMNET++ architecture for the first and second scenarios is
if the packet is lost. The loss of the packages can also take
illustrated in Figure 2. The high level VoLTE network is
place by removing all those packets that do not arrive to the
composed by the following elements:
end of the receiver on time. It becomes problematic whenever
the loss of packet takes place in a bulk. The highest rate of  Two User Equipment Support Voice Over LTE
packet loss so the voice can be heard with enough quality must
 Two eNodeB
be 1%.
The following table shows the average and the ideal quality  Four Routers
values for the Packet Loss during Voice over LTE session:
 Two S-GW
TABLE III. PACKET LOSS RATE STANDARD  One P-GW
Packet Loss Rate Quality  One Internet Host
<1% Ideal

<5% Average

4) Jitter :
Jitter is the variation in the latency of the voice packets sent
from the source to the destination. This basically occurs due to
the congestion in the network. These similar cases can be
The following Figures have been obtained after running the
simulation of the first scenario.
- MOS
The MOS of the first scenario stayed above 4 during the
simulation which falls under the category ranging between the
scale of good and excellent. The average of MOS we obtained
is 4.36, which is the normal value of any VoLTE service.
The following Figure shows the MOS obtained after
running the simulation of Scenario 1:

Fig. 2. OMNeT++ LTE Network Topology (Scenarios 1&2)

2) Scenario (1 & 2) Voice Over LTE network.


The third scenario of the VoLTE network is composed with
the help of below mentioned elements:
 Six User Equipment
 Two eNodeB Fig. 4. Scenario 1 MOS

 Four Routers - End to End Delay


 Two S-GW As we can see in figure 5, we have delay for about 8 secs
between the period 7s-11s. The average end to end delay is
 One P-GW 1.77ms which meets the standard since it’s below 50ms.
 One Internet Host

Fig. 5. Scenario 1 End to End Delay

- Packet Loss Rate


According to the explanation and description of the packet
loss, the average value obtained from the simulation is 0.21 %.
This rate is very small as compared to that of the ideal value
(1%).
Fig. 3. OMNET++ LTE Network Topology (Scenario 3)

D. Simulation Analysis & Result


This section presents the simulation analysis and result for
the conducted experiment. The quality of service can be
measured by several of the factors. In this experiment, the
quality of the network for each scenario has been compared in
terms of MOS, End to End Delay, Packet Loss Rate and Jitter.
1) Scenario 1
This scenario has been implemented to conduct an
evaluation analysis of the performance of VoLTE between two Fig. 6. Scenario 1 Packet Loss Rate
UEs. The speed of the sender and receiver of the voice is 0 m/s.
- Jitter
The jitter as seen in Figure. 7 remained static and didn’t
change over the time. It stayed at the same value (8ms) till the
end of the call. Since the result is less than 20ms this means
that the quality of voice in this scenario was excellent.

Fig. 9. Scenario 2 End to End Delay

- Packet Loss Rate


The average percentage of the packets loss for scenario 2 is
Fig. 7. Scenario 1 Jitter 7.86%. This value is more than the ideal (1%) and the average
(5%) percentage of the acceptable loss in VoLTE service. This
2) Scenario 2 variance in the percentage may affect the quality of the voice
In this scenario, we studied the evaluation of the between the sender and the receiver.
performance of VoLTE between the sender (UE1) and receiver
(UE2). The simulation was conducted for about 20s and the From this result we can conclude that the speed of UEs
speed of the UEs during the voice conversation was 100 km/h during the VoLTE session possess the capacity of directly
(28 m/s) affecting the performance of the call.

The following Figures are obtained after running the


simulation of the second scenario:
- MOS
As the following figure shows, the MOS for the second
scenarios is varying over the time. It started with 4.4, and then
dropped to 1.7, then again increased to 4.4 and finally ended up
with 1.7. The average value is 3.35 and based on the standard
rating, it can be determined that the quality of the voice is
ranging between fair and good.
Fig. 10. Scenario 2 Packet Loss Rate
- Jitter
The following figure shows the jitter result for the second
scenario. As we can see from the chart, the jitter started to
increase after 4 seconds from the start of the call. It increased
from 8ms to 23ms and then dropped to 3ms. The average jitter
we got is 7.7ms, which is acceptable as per the standard
grading.
From the result we can conclude that making VoLTE call
Fig. 8. Scenario 2 MOS
while driving may affect the quality of voice depending on the
speed of the UEs.
- End to End Delay
As we can see in the Figure 4.9, the delay started with 0ms
then after 4 seconds from the beginning of the conversation it
reached 8ms then again it decreased to 0ms. The average delay
we got in this term is 2.6ms, which is acceptable since it is less
than 50ms.

Fig. 11. Scenario 2 Jitter


3) Scenario 3
In this scenario, the sender (UE1) is calling the receiver
(UE2) via VoLTE service while other four UEs are
downloading a video of size 200mb from the server. The other
four UEs are connected to the same eNodeB as UE1.
- MOS
The MOS of scenario 3 does not have a big difference
when it is compared to the first Scenario. The MOS average
value we obtained after conducting the simulation is 4.192,
which can be considered to be close to the value of scenario 1
(4.3619).
Fig. 14. Scenario 3 Packet Loss

- Jitter
The below mentioned diagram represent the jitter result
from the third scenario. It can be seen that the jitter started to
increase after 3sec from the start of the call and again
decreased to 1ms after 17 sec. By 17 sec, it again increased to
7ms and by 20 sec, it again declined to 1ms.

Fig. 12. Scenario 3 MOS

- End to End Delay


The average delay value obtained from the simulation of
scenario 3 is 2.5ms which is much less than the ideal value
(50ms) and the average value (150ms). This means that having
a congested eNodeB while making a call in LTE network
should not lead to a big delay in the packets sent.
Fig. 15. Scenario 3 Jitter

4) Scenarios Result Side by Side


In this section, a brief comparison between the three
scenarios based on the average QoS parameters values has
been introduced.
The following table and charts compare the performance of
the three conducted scenarios

TABLE V. SCENARIOS COMPARISON


Fig. 13. Scenario 3 End to End delay QoS
Scenarios
No. Packet
MOS End to End Delay Jitter
- Packet Loss Rate Loss
Scenario 1 4.3619 1.7 ms 0.21 % 8 ms
The line chart in figure 14 shows the packet loss during the
simulation of scenario 3. As we can see the average percentage Scenario 2 3.3489 2.6 ms 7.86% 7.7 ms
of the packet loss are less than 1% of the total packets sent 4.1942 2.5 ms 0.96 % 2.8 ms
Scenario 3
which is acceptable.
- MOS

In the below diagram, the comparison of the voice call over


LTE has been compared by keeping into consideration the
three of the scenarios. Three of the different scenarios are
displaying the irregular frequency of the VoLTE MOS vector.
- Jitter

Fig. 16. Scenarios MOS

Fig. 19. Scenarios Jitter


- End to End Delay
Similar to that of the above diagrams, in this, the
comparison of the jitter results of all the three scenarios have
been mentioned. The second scenario’s jitter result show that it
is increasing from 4 sec and again at 20 sec, it declined. In the
first scenario, the jitter result is seen to be stable throughout the
call. Finally, the third scenario declines initially and then rises
to certain point and again declines drastically.

V. CONCOLUSIONS
In this paper, performance analysis of Voice over LTE
network is presented by studying the quality of service based
Fig. 17. Scenarios End to End Delay
on four of the major factors such as MOS, End to End Delay,
Packet Loss Rate and Jitter. The simulation is designed and
The average delay value of the three scenarios has been implemented with major simulation tools of OMNeT++ 4.6,
compared in the above diagram. It can be observed that the first INET Framework 2.6 and SimuLTE. Based on the simulated
scenario is increasing at 7 sec an then again declining to o ms scenarios, we found that the speed of the sender and Receivers
at 11 sec. The second scenario shows that at 4 sec, it is (UEs) are the crucial motivators that possess the capacity of
increasing and again it declines at slow rate. The final third seriously affecting the quality of the call. Once, the speed of
scenario is showing a fluctuating rate with both increase and the UEs is changed from 0 to 100km/s, the average value of
decline. MOS has been dropped from 4.3619 to 3.3489.
The jitter and the packet loss percentage were also affected
- Packet Loss
by the speed of UEs. Therefore, taking into consideration, all
these facts and other measures, the simulation module of the
LTE network with the help of OMNeT++, INET Framework
and SimuLTE has been successfully done for this particular
study. Apart from the above mentioned information, this
particular section explains the operation of the LTE network
under the variety of scenarios. Each of the scenarios explains
the performance of data and voice under the different
configurations. In all the scenarios, the description of the
performance of voice has been explained according to the
increase in the times of the general response due to the increase
in the demand for traffic along with maximum bit rate from the
different users and the maximum throughput.

Fig. 18. Scenarios Packet Loss

In the above diagram, the comparison of the packet loss REFERENCES


during the simulation of three of the scenarios is demonstrated.
The diagram shows a fluctuating rate in the different scenarios. [1] Poikselkä, M., & Et Al., 2012. Voice over LTE (VoLTE). John Wiley &
Thus, if there is a fluctuation in the rate of packet loss, it might Sons.
have adequate effect on the quality of voice between the sender [2] Roessel, S., & Et Al., 2014. Radio Network Evolution towards LTE-
Advanced and Beyond. Intel Technology Journal, Vol. 18, Issue 1,
and receiver. pp.204-227.
[3] Korowajczuk, L., 2011. LTE, WiMAX and WLAN Network Design, calculate-the-mos-score-for-qos-sensors Accessed April 02,
Optimization and Performance Analysis. John Wiley & Sons. 2016
[4] OpenSim Ltd., 2015. OMNeT++. OMNeT++ 5.0rc released. [Online] [7] Voip Think, 2015. QoS - Quality of Service. Jitter. [Online] Available
Available at: https://omnetpp.org/ Accessed April 02, 2016. at: http://www.en.voipforo.com/QoS/QoS_Jitter.php
[5] Virdis, A., Stea, G., & Nardini, G., 2014. SimuLTE – A Modular Accessed April 02, 2016.
System-level Simulator for LTE/LTE-A Networks based on OMNeT++. [8] Virdis, A., & Nardini, G., 2015. SimuLTE. What is SimuLTE: simulator
Simultech, pp. 1-11. for LTE networks? [Online] Available at: http://simulte.com/ Accessed
[6] Paessler, 2016. How does PRTG calculate the MOS score for QoS April 02, 2016.
sensors? Knowledge Base. [Online] Available at: [9] Breitenecker, F., & Troch, I., 2015. Simulation Software-Development
https://kb.paessler.com/en/topic/59491-how-does-prtg- and Trends. Control Systems, Robotics and Automation, Vol. 4, pp. 1-13

View publication stats

You might also like