OPTIMIZING VOIP USING A CROSS LAYER CALL ADMISSION CONTROL SCHEME

International Journal of Computer Networks & Communications (IJCNC) Vol.5, No.4, July 2013
DOI : 10.5121/ijcnc.2013.5410 117
OPTIMIZING VOIP USING A CROSS LAYER CALL
ADMISSION CONTROL SCHEME
Mumtaz AL-Mukhtar and Huda Abdulwahed
Department of Information Engineering, AL-Nahrain University, Baghdad, Iraq
almukhtar@fulbrightmail.org
ABSTRACT
Deploying wireless campus network becomes popular in many world universities for the services that are
provided. However, it suffers from different issues such as low VoIP network capacity, network congestion
effect on VoIP QoS and WLAN multi rate issue due to link adaptation technique. In this paper a cross layer
call admission control (CCAC) scheme is proposed to reduce the effects of these problems on VoWLAN
based on monitoring RTCPRR (Real Time Control Protocol Receiver Report) that provides the QoS level
for VoIP and monitoring the MAC layer for any change in the data rate. If the QoS level degrades due to
one of the aforementioned reasons, a considerable change in the packet size or the codec type will be the
solution. A wireless campus network is simulated using OPNET 14.5 modeler and many scenarios are
modeled to improve this proposed scheme.
KEYWORDS
VoIP Capacity, QoS, Cross-layering, VoWLAN, Codec Adaptation.
1. INTRODUCTION
Voice over IP (VoIP) has been widely used these years for its simplified infrastructure and
significant cost savings. One of the most interesting use cases for VoIP is in combination with the
IEEE 802.11 technology to provide wireless voice services to mobile devices such as laptops,
smart phones and PDAs. The use of VoWLAN (VoIP over WLAN) makes it possible for mobile
employee of an enterprise or a campus to be provided with cost effective voice and flexible
services [1]. However, two technical problems need to be solved. The first is that the system
capacity for voice can be quite low in WLAN. The second is that VoIP traffic and data traffic
from traditional applications (web, e-mail, etc...) can interfere with each other and bring down
VoIP performance.
Voice capacity is defined as the maximum number of voice sessions that can be supported
simultaneously by a network under specific quality constraints [2]. Therefore, it is essential to
determine the number of users a WLAN can support simultaneously without significantly
degrading the QoS and to analyze the delay, jitter and packet loss of VoIP over WLAN.
The objective of this paper is to study the effect of different codecs in IEEE 802.11 multi-rate
environment, and the VoIP packet payload size in order to develop a cross-layer call admission
control scheme between MAC and application layers. This mainly aims to enhance the network
capacity and quality for the VoIP calls with the accepted QoS constraints. This research focuses
on addressing the congestion in the network and the multi-rate issue caused by the link adaptation
technique in WLAN campus network.
The rest of the paper is organized as follows: section 2 introduces VoIP system, VoIP quality
evaluation criteria, and related link adaptation concept. Section 3 provides a brief overview of
several researches in the related area. Section 4 introduces the cross-layer call admission control

118
scheme. The simulation scenarios and results are presented in section 5. Finally, conclusion is
drawn in section 6.
2. TECHNICAL BACKGROUND
This section gives a VoWLAN overview inspecting VoIP quality constraints. Our focus in this
research is to achieve better QoS under varying network conditions.
2.1. VoIP System
VoIP system consists of three essential components: codec, packetizer and playout buffer. At the
sender side, the analog voice signals are converted to digital signals, compressed and then
encoded by voice codecs. There are various voice codecs that are developed and standardized by
the International Telecommunication Union - Telecommunication Sector (ITU-T), such as G.711,
G.726, G.729, G.723.1a, etc. The subsequent process performed is packetization, where the
encoded voice is fragmented into equal size of packets by the packetizer. Each packet generated is
composed of the encoded voice and headers, which are added at different layers, specifically by
layers of Real-time Transport Protocol (RTP), User Datagram Protocol (UDP), and Internet
Protocol (IP), as well as by Data Link layer header. Moreover, RTP and Real-Time Control
Protocol (RTCP) are designed at the application layer to support real-time applications [3].
The packets are sent out over IP network to its destination, where the reverse process of decoding
and depacketizing of the received packets is carried out. Additionally, there are signaling
protocols of VoIP namely Session Initiation Protocol (SIP) and H.323. These protocols establish
VoIP calls. H.323 was standardized by ITU-T specifically to smoothly work together with PSTN.
On the other hand, SIP was standardized by Internet Engineering Task Force (IETF) to support
Internet applications, such as telephony. Figure 1 illustrates VoIP protocol stack with respect to
its TCP/IP protocol standard [4].
Figure 1. VoIP implementation in TCP/IP protocol standard
2.2. VoIP QoS Evaluation
The voice can be tested for quality in two ways, namely, subjective and objective. Humans
perform the subjective voice testing by listening to the voice sample, whereas, objective tests are
performed by computers. Common subjective benchmark for quantifying the performance of the
speech codec is the Mean Opinion Score (MOS). For performing MOS test, a voice sample is
given to a group of listeners. They listen to the sample and give a rating on a scale where

119
"excellent" quality is given a score of 5, "good" a 4, "fair" a 3, "poor" a 2, and "'bad" a 1. The
ratings given by every member of the group is then averaged to get the MOS. The E-Model is
most commonly used for objective measurements. The basic result of the E-Model is the
calculation of the R-Factor. The R-factor is defined in terms of several parameters associated with
a voice channel across a mixed Switched Circuit Network and a Packet Switched Network. The
parameters included in the computation of the R factor are fairly extensive covering such factors
as echo, background noise, signal loss, codec impairments, and others. R-factor can be expressed
by (1) [5].
R = R0 + Is + Id + Ie + A (1)
Where, R0 groups the effects of noise, Is includes the effect of other impairments related to the
quantization of the voice signal, Id represents the impairment caused by delay, Ie covers the
impairments caused by low bit rate codecs and packet losses. The advantage factor A
compensates for the above impairments under various user conditions. For mobile telephony A is
assumed to be 10. We consider that A is 0 in the case of VoIP.
MOS is related to R-Factor by (2) [2].
For R<0: MOS=1
For R>100: MOS=4.5
For 0<R<100: MOS= 1+0.035R+7×10−6
R (R−60) (100−R) (2)
The voice performance is considered acceptable if the end-to-end delay is less than 150ms and the
packet loss rate is less than 2% [6].
2.3. IEEE 802.11 Standards and Link Adaptation
A variety of wireless LAN technologies exist that use various frequencies, modulation techniques,
and data rates. IEEE provides most of the standards, which are categorized as IEEE
802.11,802.11a,802.11b,802.11g,802.11e,…. ,and 802.11n. For instance 802.11b operates at 1, 2,
5.5, 11 Mbps, 802.11a & 802.11g can support up to 54 Mbps [7]. IEEE 802.11 supports the link
adaptation technique, which allows a wireless transmitter to select an appropriate transmission
rate on a packet basis according to the wireless link conditions. However, if a lower transmission
rate is selected because of deterioration of the wireless link conditions, packet delays and losses
will be increased due to the poor bandwidth availability, especially when the traffic load is high,
and as a result, the quality of real-time applications will significantly deteriorate. The IEEE
802.11 standard does not specify how and when to switch between the permitted rates, so various
link adaptation (LA) mechanisms have been studied [8]. One of the most effective solutions for
this issue is to adapt the codecs of some of the active voice flows using codec adaptation
algorithm (CAA) [9]. Cross layer methods have been designed to overcome numerous setbacks
that were previously faced by wireless applications running on the strict layered OSI protocol
stack. Cross layer design has numerous attributes. These include: fast stack adaptation, cross-
protocol coordination, ease of QoS provisioning to different applications, prioritization ease, no
unnecessary interaction between layers and improved throughput [10].
3. RELATED WORK
The two important issues related to VoWLAN are capacity and QoS depend on different
parameters such as data rate, speech codec, delay, jitter and packet loss. Several techniques and
algorithms have been presented to improve the quality of VoIP over WLAN. Cross-layer

120
interaction approach has been introduced in several possible ways to optimize different layers’
parameters. Papapanagiotoua et al. [11] exploit the interrelations between data rate, packetization
interval, packet error rate, and retransmission attempts to produce a model to be used in the cross
layer call admission control scheme in order to optimize QoS for single cell WLAN.
Several researches [12, 13, 14, 15, 16] propose techniques to mitigate the effect of multi rate
issues on VoIP quality over WLAN by changing either packet size or codec type depending on
RTCP receiving messages and MAC layer data rate using either a cross layer control scheme or
an adaptive algorithm. Luthra and Sharma [17] suggest VoWLAN QoS enhancement using
numbers of QoS techniques like Integrated Services, Differentiated Services, and Resource
Reservation Protocol. While Chakraborty et al. [18] introduce QoS upgrading by optimizing the
concerned Access Points' parameters such as buffer size, retransmission limit, RTS threshold,
transmission power, antenna type, location factors and network load. In [19] IEEE 802.11e
standard has been harnessed to enhance the QoS compared with standard 802.11b/g.
The QoS for VoWLAN can be guaranteed using the smart call admission control scheme that is
based on a dynamic bandwidth channel allocation to reserve the greater BW for VoIP application
as introduced in [20]. However, authors in [21] and [22] invest for this purpose CAC (Call
Admission Control) based on channel load estimation that is built around the TBIT (Time
Between Idle Times) scheme. Jung et al. [23] employ call admission control scheme to achieve
high link utilization via investigation of the on-off patterns of VoIP traffic from the Brady model.
Chang et al. [24] propose a QoS – aware path switching strategy by using stream control
transmission protocol (SDCTP) in Multi–Protocol Label Switch (MPLS) network to improve
VoIP traffic. The main contribution of this paper is that, the proposed CCAC scheme will
enhance both the capacity and the QoS for VoWLAN in a campus-congested network based on
two parameters: packet size and codec type.
4. THE PROPOSED CROSS LAYER CALL ADMISSION CONTROL SCHEME
In order to improve the capacity and the QoS of VoIP over a WLAN, a cross layer call admission
control scheme (CCAC) is proposed to allow the communication between MAC and application
layers. This is depicted in figure 2.
Figure 2. CCAC proposed scheme
This scheme tries to increase network VoIP capacity and to reduce the effect of two problems: the
congestion and the WLAN link adaptation technique on VoIP calls. The periodic monitoring of
the RTCPRR (Real Time Control Protocol Receiving Report) messages will be helpful to find if
there is congestion in the network by evaluating the R-factor from the values of the end to end
delay and packet loss ratio parameters of these messages. If these values are greater than the
threshold values so that the end to end delay >150 ms and packet loss ratio >2% then the QoS for

121
VoIP will be degraded indicating that R<70 which is minimum value for the accepted QoS. In
this case, the VoIP packet size will be checked whether it exceeds the predefined maximum size
for the negotiated codec. In negative case the packet size will be increased by one frame and the
monitoring process is resumed. However, if packet size exceeds the maximum then the codec will
be changed to the one of a lower bit rate. Table 1 illustrates parameters of different codecs. If the
transmission rate changes due to link adaptation algorithm then an alarm is send from the MAC
Layer to the CCAC scheme which also decides to either adjust the packet size or change the
codec type according to the RTCPRR. A descriptive flowchart of CCAC functions is shown in
figure 3.
Table 1. CODEC Parameters
CODEC Type G.711 G.726 G.729A G.723.1
Bit rate kbps 64 32 8 6.3
Bits per frame 8 4 80 159
Algorithmic delay (ms) 0.125 0.125 15 37.5
Codec delay (ms) 0.25 0.25 25 67.5
Compression type PCM ADPCM CSACELP ACELP
Complexity (MIPS) << 1 ≈ 1 <= 11 <=18
MOS 4.1 3.85 3.7 3.6

122
Figure 3. CCAC Flow chart
5. SIMULATION & RESULTS
To assess how the proposed CCAC scheme can enhance the network performance in different
conditions, the university wireless campus network has been simulated using OPNET modeler
14.5 [25]. Many scenarios have been imposed using different packet size, data rate, and different
codecs to cover all possible cases.
5.1. Simulation Setup
The simulated wireless campus network consists of seven buildings; each one has three WLAN
subnets with two servers (FTP& DB) and a separate subnet of three wireless servers dedicated for
Email, Heavy browsing & Web searching applications. In addition there is an SIP sever for calls
management. This network is connected to the Internet as shown in figures 4, 5, and 6.

123
Figure 4. Campus Network Figure 5. Building Subnet
Each WLAN subnet has 20 wireless work-station mobile nodes communicating with each other
through an Access Point. The nodes are using IEEE802.11b standard for their communication
with transmission rate of 11 Mbps. Six applications are defined in this network which are VoIP,
FTP, DB, Email, web searching, and heavy browsing. The codec chosen for VoIP is G.711 with
one frame per packet. The VoIP call is assumed to be between two nodes, caller and the callee.
Figure 6. Server's Subnet
5.2. Simulation Scenarios and Results
Simulation scenarios cover two main cases as follows:
5.2.1. Effects of Congestion and Addition of New VoIP User
The first scenario aims to find how many simultaneous calls can be achieved in the network with
packet size of one frame and G.711codec for VoIP application. Only five calls can be found in the
same time so that the required end to end delay, MOS, and packet loss will be acceptable. Adding
new calls or other traffic cause the performance to be unacceptable. These results are depicted in

124
figures 7 and 8. In the second scenario, the VoIP packet size has been increased to comprise two
frames. In this scenario, 10 simultaneous VoIP calls can be found and other traffic could be
added. In the third scenario, the packet size has been increased to 3 frames to achieve 14 calls
with more added traffic. In the fourth scenario the number of frames (F) considered is 4 and the
simultaneous calls (N) are 18. In the fifth scenario F=5 and N=64 with each node in the network
has at least three running application so that the network is congested. The MOS is found to be
>3, end to end delay <150 s and packet loss ratio <2% which are accepted. Adding more frames
to the VoIP packet of the G.711 codec will produce unacceptable results. Figures 9a& b show the
maximum number of N and MOS values with respect to F. Next scenario is considered when
there is a new call, a codec change for this call to G.726 of 32 Kbps with 2 frames per packet will
occur for this call to be accepted.
Figure 7. End to End Delay (case1, scenario1)
Figure 8. Traffic send and Traffic received (case1, scenario1)

125
Figure 9a. Max. N with respect to F Figure 9b. MOS value with respect to F
5.2.2. Effect of Multi Rate Issue
Many scenarios have been introduced to cover changes in transmission rate from 11 to 5.5, 2, and
1 Mbps for certain VoIP nodes. The packet size has been varied from 1 to 5 frames and the codec
deployed is G.711. In the first scenario, the VoIP packet contains one frame and the transmission
rate varies to 5.5 Mbps. Only the end-to-end delay will be affected but it is still accepted.
Nevertheless, when a rate of 2 & 1 Mbps is deployed, the performance will be unacceptable. By
increasing the packet size to 2 frames the performance is enhanced as shown in figures 10 and 11.
Figure 10. End to End Delay (case 2, scenario 2)
Figure 11. MOS value (case 2, scenario 1)

126
The second scenario is considered when the VoIP node’s transmission rate is 11Mbps but the
packet size is 2 frames and a transmission rate changes within 5.5, 2, and 1 Mbps. A change in the
packet size is made to enhance the VoIP quality of service parameters so it can be accepted. This
is shown in figures 12 & 13. When the transmission rate is 2 Mbps and the frame size is 2 frames
the end to end delay is not acceptable i.e. > 150 ms. However, at a packet size of 3 frames the end
to end delay will be acceptable.
Figure 12. End to End Delay (case 2, scenario2)
The third scenario is considered when VoIP packet size is 3 frames at transmission rate of 11
Mbps and it varies to 5.5, 2, and 1Mbps. At 5.5& 2 Mbps, the packet size changes to 4 frames to
get the accepted QoS parameters values. At 1 Mbps increasing the packet size will not enhance
the performance so the codec type will be switched to G.726 with packet size of 4 frames as
shown in figures 14&15.

127
Figure 14. End to End delay (case 2, scenario 3)
The fourth scenario is considered when the VoIP packet size consists of 4 frames and some VoIP
nodes change their transmission rate from 11 Mbps to 5.5, 2, 1 Mbps. At 5.5Mbps changing the
codec will be more effective than packet size so G.726 will be the codec. At 2 Mbps also the
codec will be changed to the one of a lower bit rate which is G.729 A. That is even with 5 frames
packet size of the G.726 codec the quality would not be acceptable. At 1Mbps G.729A with 5
VoIP frames also does not provide good quality so it is replaced with G.723.1 of 6.3 Kbps so the
calls could be continued. These results are depicted in figures 16&17.

128
Figure 16. End to End delay (case 2, scenario 4)
6. CONCLUSION
In this paper, the VoWLAN application is tested for performance in campus network. A cross
layer call admission control scheme is proposed to enhance network capacity and QoS for VoIP
by adapting packet size or codec type. The simulation results proved that the proposed CCAC
scheme provides an efficient and fast way for solving the network congestion effect on VoIP
application. Increasing network VoIP capacity and reducing the multi rate effect is achieved by
monitoring the RTCPRR for the last end to end delay and Packet loss ratio that provides the
network QoS level, which is represented by R-factor. As well as the MAC layer is monitored for
any changes in the data rate of the mobile node. If the QoS degrades due to one of the above
problems then the solution will be by either varying the packet size or codec type to optimize
network performance without dropping the call.

129
REFERENCES
[1] X. Wei, Y. Bouslimani, & K. Sellal, (2012) “VoIP Based Solution for the Use over a Campus
Environment” 25th IEEE Canadian Conference on Electrical and Computer Engineering
(CCECE), PP 1-5.
[2] C. Brouzioutis , V. Vitsas & P. Chatzimisios, (2010) “Studying the Impact of Data Traffic on
Voice Capacity in IEEE 802.11 WLANs”, IEEE International Conference on Communications, PP
1-6.
[3] M. N. Ismail, (2010) “Analysis of Secure Real Time Transport Protocol on VoIP over Wireless
LAN in Campus Environment” International Journal on Computer Science and Engineering
(IJCSE), Volume 02, No. 03, PP 898-902.
[4] S. A. A. Alshakhsi & H. Hasbullah, (2010) “Improving QoS of VoWLAN via Cross-Layer
Interaction Approach”, International Symposium on Information Technology, PP 678-682.
[5] A. Mukhopadhyay, T. Chakraborty, S. Bhunia, I. Misra & S. Sanyal, (2011) “Study of Enhanced
VoIP Performance under Congested Wireless Network Scenarios”, Third International Conference
on Communication Systems and Networks (COMSNETS 2011), PP 1-7.
[6] W. Wang, S. Liew, & V. Li, (2005) “Solutions to Performance Problems in VoIP Over a 802.11
Wireless LAN”, IEEE Transactions on Vehicular Technology journal, Volume 54, Issue 1, PP
366-384.
[7] D. E. COMER, (2008) Computer Networks and Internets, Pearson Education, Inc., Fifth Edition.
[8] T. Kawata, & H. Yamada, (2006) “Adaptive Multi-Rate VoIP for IEEE 802.11 Wireless Networks
with Link Adaptation Function”, Proceedings of the Global Telecommunications Conference
(GLOBECOM '06), PP 1-5.
[9] Z. Chen, L. Wang, F. Zhang, X. Wang & W. Chen, (2008) “VoIP over WLANs by Adapting
Transmitting Interval and Call Admission Control”, IEEE International Conference on
Communications, PP 3242-3246.
[10] V. Sentongo & H. Chan, (2009) “Optimization of Quality of Service Requirements for Real-Time
Applications using Cross Layer Design”, IEEE AFRICON'09, PP 1-9.
[11] I. Papapanagiotoua, F. Granellib, D. Kliazovichb, & M. Devetsikiotis, (2011) “A Metamodeling
Approach for Cross-Layer Optimization: A Framework and Application to Voice over WiFi”,
Simulation Modelling Practice and Theory Journal, Volume 19, Issue 9, PP 2117-2129.
[12] S. Ali Alshakhsi, & H. Hasbullah, (2011) “Improving QoS of VoWLAN via Cross-Layer-Based
Adaptive Approach”, International Conference on Information Science and Applications, PP 1-8.
[13] M. Tuysus & H. Mantar, (2010) “Evaluation of Cross Layer QoS Approach for Improving Voice
Quality over Multi Rate WLANs”, International Conference on Computer Engineering and
Systems (ICCES), PP 73-78.
[14] A. Sfairopoulou, B. Bellalta, C. Macian, & C. Oliver, (2011) “A Comparative Survey of Adaptive
Codec Solutions for VoIP over Multirate WLANs: A Capacity versus Quality Performance Trade-
Off” EURASIP Journal on Wireless Communications and Networking, Volume 2011, Issue 1, PP
1-13.
[15] H. Kazemitabar & A. Saida, (2011) “An Adaptive Rate Control Algorithm for VoIP over Multi-
Rate WLANs”, 2nd World Conference on Information Technology (WCIT-2011), PP 1087-1092.
[16] P. McGovern, P. Perry & S. Murphy, (2011) “Endpoint-Based Call Admission Control and
Resource Management for VoWLAN”, IEEE Transactions on Mobile Computing, Volume 10,
NO. 5, PP 684-699.
[17] P. Luthra & M. Sharma, (2012) “Performance Evaluation of Audio Codecs using VoIP Traffic in
Wireless LAN using RSVP”, International Journal of Computer Applications, Volume 40, No.7,
PP 15-21.
[18] T. Chakraborty, A. Mukhopadhyay, S. Bhunia, I. S. Misra & S. K. Sanyal, (2012) “An
Optimization Technique for Improved VoIP Performance over Wireless LAN”, Journal of
Networks, Volume 7, NO. 3, PP 480-493.
[19] S. V. Bhanu & R. M. Chandrasekaran, (2012) “Voice Call Capacity Analysis and Enhancement of
IEEE 802.11 WLAN”, European Journal of Scientific Research, Volume 76, No.2, PP 271-280.
[20] R. Ganiga, B. Muniyal & Pradeep, (2012) “ Characteristic Analysis of VoIP Traffic for Wireless
Networks In Comparison with CBR using QualNet Network Simulator” International Journal of
Computer Applications, Volume 50, No.11, PP 25-31.

130
[21] K. Yasukawa, A. G. Forte & H. Schulzrinne, (2007) “Distributed Delay Estimation and Call
Admission Control in IEEE 802.11 WLANs”, IEEE International Conference on Network
Protocols, PP 334-335.
[22] P. Dini, N. Baldo & J. Nin-Guerrero, (2010) “Distributed Call Admission Control for VoIP over
802.11 WLANs based on Channel Load Estimation” IEEE International Conference on
Communications journal, PP 1-6.
[23] Jung Ji-Young , Seo Dong-Yoon & Lee Jung-Ryun, (2013) "VoIP Call Admission Control
Scheme Considering Voip on-off Patterns", International Conference on Information Networking
(ICOIN), PP 371 – 374.
[24] Chang Lin-huang, Lee Tsung-Han, Chu Hung-Chi, Lo Yu-Lung & Chen Yu-Jen, (2013) "QoS-
Aware Path Switching for VoIP Traffic using SCTP", Computer Standards & Interfaces Journal,
Volume 35, Issue 1, PP 158–169.
[25] A. S. Sethi & V.Y. Hnatyshin, (2013) The Practical OPNET User Guide for Computer Network
Simulation, CRC Press.

OPTIMIZING VOIP USING A CROSS LAYER CALL ADMISSION CONTROL SCHEME

More Related Content

What's hot (19)

Viewers also liked (19)

Similar to OPTIMIZING VOIP USING A CROSS LAYER CALL ADMISSION CONTROL SCHEME (20)

More from IJCNCJournal (20)

OPTIMIZING VOIP USING A CROSS LAYER CALL ADMISSION CONTROL SCHEME