VoIP Quality of Service Evaluation.

Executive Summary

This research paper explores the performance of VoIP, as well as the parameters that affect the quality of service of VoIP. The different communication aspects of VoIP namely; call signaling protocols and networking environments have also been explored, with the latter’s emphasis on LAN and WAN. Performance evaluation is about identifying QoS parameters that would be of relevance to VoIP communications. Consequently, these parameters have been measured relative to RTP packets transmission in the communication of VoIP, with the obtained results showing the significant impact the assessed communication aspects, on the QoS in VoIP communication. Ultimately, the impact shall vary depending on the selected communication aspects for analysis.


As a technology, voice over internet protocol (VoIP) allows user to make calls with the aid of a broadband internet connection; as opposed to the use of the analogue lines. In addition to the offering of video and data services, cable operators have now shifted focus offering subscribers VOIP, in a bid to prevent an imminent defection to networks now offering satellite services.

With the advent of Voice over Internet Protocol, the telephony world has witnessed a tremendous change in terms of technology, a change that is about to witness the phasing out of the traditional PSTN network (Miloucheva 2004).

With this evolution accelerating everyday at an alarming rate, it is only wise then to thoroughly study what VoIP can offer, in its quest to replace the traditional PSTN network. In the implementation of VOIP, Quality of Service (QoS) becomes a major issue. In this regard, it becomes important that one is able to ensure that the packet traffic for a voice is guaranteed of not being delayed at best or dropped at worst, following interferences as a result of low traffic (Kos, Klepec & Tomažič 2002).

To ensure such a guarantee then, the latency, packet loss, jitter and burstness of loss and jitter, ought to be considered. For an end user, bad echoes can result from large delays, thus making a working conversation difficult to attain, owing to interruptions. Jitters (variations in delay of packet delivery) have the potential to cause strange sound effects. However, this can be overcome by the inclusion of ‘jitter buffers’ in the software (Han, Ahn, & Chung 2002).

On the other hand, interrupts can come about due to packet loss (too much traffic in the network that leads to the network dropping packets). However, only lots of packets loss are capable of making lousy noises. For anyone migrating to internet telephony, or even using a VOIP network, VOIP QoS then become an important consideration. Indeed, this is the perfect combination that has been missing in the telephone network. In comparison with the traditional public switched telephone network (PSTN), VOIP has the potential of delivering a higher QoS (Jha & Hassan 2002).

Other than offering users multivariate features, VOIP also enables them to realize cost savings, in comparison to the traditional PSTN. In terms of carrier lines, PSTN requires dedicated lines from the telephone company; while with VOIP, all one needs is an internet connection to transmit all voice channels. For bandwidth, each analogue telephone line only uses about 64kbps per direction (Kos et al 2002). On the other hand, VOIP, with the aid of compression, could as well make use of as little as 10 kbps per each direction. For a PSTN, call waiting, music on hold, conferencing, caller ID and other features, are often made available at a fee, while these are made free with a VOIP network.

In terms of expandability and upgradeability, PSTN is both complex and expensive, while VOIP only requires extra internet bandwidth, as well as a software upgrade (Miloucheva 2004). Through the help of VOIP technology, it is now possible to transmit human voices on an IP network, to any corner of the world. This then ensures that long distant calls are now more affordable. Given that VoIP is a real time transmission, its management than becomes paramount.

Aims of the project

This project is aimed at helping in the understanding how VoIP and QoS works, as well as in the assessment of the parameters of QoS. Further, the research is also aimed at evaluating how QoS is able to perform under WAN and LAN architectural networks of VoIP, while at the same time comparing these to a signaling protocol.

Project objectives

  • Focusing on the study of the inherent working nature of the VoIP Quality of Service in WAN and LAN network environments.
  • Evaluating the Performance of signaling protocols such as H.323 and SIP in VoIP quality considerations.
  • Evaluating and understanding the parameters for VoIP quality of service.
  • Determining the basic and complete working of parameters involved in VoIP Quality of Service will be studied and understood.
  • Assessing the behavior of different VoIP quality parameters.


For a network that is either to internet telephony, or one already using VoIP, QoS in VoIP is significantly important. For a long time now, use of telephone QoS has been always taken for granted. However, those networks that are able to switch to the VoIP service enter into a field whereby the provision of quality service is both elusive and hard to achieve (Hole & Tobagi 2002). The field is dogged by latency, network jitter, and packet loss, often leading to a distorted VoIP transmission, with the possibility of a total collapse.

Despite these shortcomings, VoIP still has an enormous potential in terms of delivering quality service, when compared to the traditional public switched telephone network (PSTN). Furthermore, an accurate organizing of both bandwidth and the ensuring of the fact that costs are low, then VoIP has the potential to achieve QoS at maximum benefits. As a form of technology, VoIP bridges the old and the new; time-division multiplexing on the one hand, and IP on the other hand (Goode 2002).

With the help of IP phones, it is possible to deploy VoIP end-to-end. Alternatively, VoIP could also be phased in over time, a strategy referred to as migration. This strategy is gaining popularity among service providers due to its potential in voice service, video as well as data provision. Moreover, VoIP can also be deployment is also capable of being configured, thus enabling it utilize WAN links (Kos 2002). This way, it has the capability of spilling over into its links with PSTN, in the event that the allocated capacity to WAN has been exhausted. Through such an arrangement, toll savings are enabled; up to the WAN allocation limit. Further more, VoIP allows for the integration of useful applications with the functionality of the telephone, using LAN as a transport mechanism.


In terms of quality of service evaluation, the concept of VoIP interconnection affects technological, political and economical areas. This dissertation work investigates the parameters of both VoIP and QoS, and the performance of the latter under either WAN or LAN networks. However, the performance of VoIP software is subject to a signaling protocol on the network, and this possibility is also explored in this study. Parameters of for VoIP quality of service shall also be addressed. At the same time, the workings of such parameters shall also be determined.

Literature review

A lot of research has been carried out in the field of telephony, and especially with regard to VoIP and QoS in VOIP. In the field on networking, VOIP is rapidly gaining a foothold, following efforts by industry players and experts to provide the end user with superior telephone services qualities, in comparison to the PSTN (Goode 2002). In order to effectively transmit voice data over the internet protocol network, there is a need to have an enhanced quality of service (QoS) in the network. (Miloucheva 2004) carried out a survey with a view to describing how voice could be managed over the IP networks. The study found out that indeed, VoIP is a viable alternative to the traditional analog telephone.

Even with the end users having embraced this technology, VoIP is very much under attack. Despite all these prevailing threats and vulnerabilities, the huge numbers of VoIP users, service interruptions and eavesdropping could be viewed at as the hallmark of the lack of privacy in telephone conversation. However, analysts and vendors have been working hard to find solutions. In most of the newer VOIP solutions, these are found in a closed system, in which packetised voice runs over LAN. In addition, it is now recommended that voice be separated from data traffic, in a bid to protect LAN from malware, Dos attacks and eavesdropping (Jha & Hassan 2002).

In this regard, the encrypting of voice-signaling data becomes vital, as are the provision of high security environments, and the management of VoIP interactions. Still, there are challenges ahead in the quest for efficient provision of VoIP services. The ultimate goal of VoIP is the offering of telephone services that are both affordable and efficient. Luckily, the internet has come to the rescue, but there is a potential security threat in the network, as a closed system cannot be realized. For one, it would be unrealistic to keep voice off data completely, while avoiding the use of softphones at the same time. According to Karam and Tobagi (2001), the interconnection between data and voice is in the converging applications that are bound to evolve with time.

The traditional voice providers are often naïve about the potential threats to VoIP, as PSTN switches are not at risk of being attacked by say, a virus (Kos 2001). Some of the security measure that vendors have adopted such as voice encrypting, have been touted as being less practical and not widely used, according to critics. They fear that in the long-term, VoIP may be plagued by the same invasions that have affected e-mail and other PC communications. However, VoIP vendors have already foreseen such a possibility, and are now trying to improve on VoIP security (Hole & Tobagi 2002).

By studying VoIP, as well as its performance levels, one would be better placed to understanding how voice traffic is treated by the system. Given that VoIP is a very sensitive service, the perceived quality of service (QoS) that the network so purports to provide, could as well be deteriorated as a result of flooding, with the possibility of crippling down the sender-receiver path (Karam &Tobagi 2001). Such paths would entail IP telephones, soft switches, and SIP proxy servers. Previously, VoIP providers were under pressure to provide quality and reliable services (Amir et al 2004).

However, the VoIP network has witnessed a tremendous improvement. According to the CEO of a leading VoIP enhanced service provider, the real power of VoIP lies in its application. David Hofstatter of CallWave, Inc. believes that VoIP has tremendously transformed over the years. No longer just a means for transport, Hofstatter reckons that the true value of VOiP lies in its ability to act as a platform for the applications that shall ultimately add content and value to a call (Miloucheva 2004). However, the industry is yet to embrace the enormous potential that has emerged following the emergence of VoIP as a consumer standard for voice communications.

In order to make VOIP services available, it is mandatory that an IP infrastructure is developed. In itself, an IP infrastructure development serves as both a challenge to the regulation of the service and a market (Jha & Hassan 2002). An efficient regulatory regime is an aid in the development and extension of IP even to remote areas. It is also possible to provide for measurable QoS in the case of a well managed VoIP service. The beauty of using VoIP is that one can be able to maintain the positioning of emergency calls, as long as the service is used from a fixed location (Kos et al 2002).


QoS Parameters

It is in the keeping with QoS for VoIP that researchers have been able to come up with QoS parameters. For this reason, this paper is focused at studying QoS network, with a reference to the QoS framework, as per the suggestions of Jha and Hassan (2004). The implication then is that the analysis shall be focused on the evaluation of the network under which VoIP communication takes place. For purposes of implementation, key QoS parameters have been identified. Table 1 in the analysis and presentation section provides these key parameters, and which were derived from Goode (2002).

Table 1: QoS Parameters

category parameters
timeliness Delay
Response time
bandwidth System- level data rate
Application-level data rate
Transaction time
reliability Mean time for failure (MTFF)
Mean time to repair (MTTR)
Mean time between failures (MTBF)
Percentage of time available
Packet loss rate
Bit error rate

For the benefit of this research analysis, three Qos parameters have been selected namely; packet loss, jitters, and delays. Given that this study is aimed at studying VoIP communication over different environments, time and reliability would thus be the major evaluation concerns. This then justifies the selection. According to results of studies conducted by Markopoulou, Tobagi and Karam (2003), high delay jitters were shown to be experienced by a large number of internet paths whose VoIP performance was poor. At the same time, the QoS parameters adopted in this research have previously been adopted in studies by Kos et al (2002), Feigin et al (2000), and Shirdokar (2001)

Related works

Previously, researchers have carried out related work in this same field. These includes Markopoulou et al (2003), Hole and Tobagi (2004), Karam and Tobagi (2001), Han, Ahn and Chung (2004), Amir, Danilov, Hedqvist and Terzis (2003), and Bearden et al (2002). Markopoulou et al (2003), undertook observations on the QoS for VoIP, while Han et al were mainly focused on the use of simulation to detect delay patterns in a VoIP network.

On the other hand, Miloucheva, Nassri, and Anzalon (2004), undertook an analysis of automated QoS parameters for the application of VoIP. Indeed, this work has been a real motivation to the undertaking of the research. Through the adoption of lab experimentation, the control of the environment becomes now more feasible; as opposed to the use of either simulation or real-life experimentation. This however, does not rule out the possibility of errors in the lab experiments, but close approximations are guaranteed.

Communication aspect

During analysis, VoIP call signaling protocols and the network environment were considered as communication aspects.

VoIP call signaling protocols

For the purpose of this research, two of the most commonly applicable call signaling protocols was selected. These are H.323 and SIP, the latter being a call signaling protocol that has been introduced by the International Telecommunication Union (ITU) (Goode 2002). SIP on the other hand is a development of the Internet Engineering Task Force (IETF), which was viewed at as an alternative to H. 323 (Goode 2002).

Network Environment

For the benefit of this study, two different network environments have been adopted namely; LAN and WAN. Combined, the two networks provide the best-effort mechanism, and in which there lacks a differentiation of the specific data types being transmitted. In essence, this is the best choice for observing VoIP performance under these conditions.


In the analysis of QoS performance for VoIP communications, this research took into accounts both secure and non-secure VoIP communications. Implementation for the VoIP communications was done with both call signaling protocols for the tests; H.323 and SIP. In order to simulate a real network behavior, a queuing mechanism had to be adopted. The results that were obtained include the values of jitter, delay, and packet loss, as obtained from each one of the VoIP implementations.

Results analysis and discussions

Table 2: Token Bucket Filters (TBF) parameters that have been adopted

Parameters Description Values selected
Bucket/burst size Size of the bucket also considered as the burst size, upon which the queue bursts once the bucket is full 1024 Kbytes
Latency Time during which a given packet resides in a TBF bucket 100ms
Rate Arrival rates for tokens 50kbps

Fidler (2003) opines that traffic shaping could as well be used as a means of reducing impacts of an interfering burst with regard to a network performance. However, higher delays in non-ideal secure inter-network connections were incurred by both H. 323 and SIP.

Table 3: Comparing Average Delays for RTP Packets transmission in SIP VoIP communication (Non-Ideal Network-to-Network)

Delay (ms) Network Conditions SIP N-N SIP PPTP SIP IPSEC
32.5 38.8 41.2

Table 4: Comparing Average Delays for RTP Packets transmission in H. 323 VoIP communication (Non-Ideal Network-to-Network)

Delay (ms) Network Conditions SIP N-N SIP PPTP SIP IPSEC
32.5 38.8 41.3

The results obtained from the delay analysis were lower, in comparison to those obtained by Markopoulou et al (2003), an assessment that involved the real internet backbone. However, these findings are considered acceptable, given that the experiment was undertaken in a controlled lab environment, and that only one VoIP communication was conducted. Amir et al (2004) argues that delay values between 100-150 ms are not only detectable by humans, but are also capable of impairing a conversation’s interactivity. Although the results obtained fall below these values, the research was nevertheless undertaken in a controlled environment, with a possibility of experiencing less severe jitter and delays in comparison with the real environment.

Table 5: A comparison of Average Jitter Values for RTP Packets of SIP and H.323 VoIP

Jitter Value (ms)
H. 323
Network conditions Network-to-Network PPTP IPSec

With respect to the analysis of packet loss, relatively high packet loss rates were consistently produced by both the non-ideal and non-ideal secure network-to-network environments. To obtain packet loss rate, the RTP packets unreachable to the destination is determined, over the number of transmitted RTP Packets. The delay distributions for both SIP and H. 323 were almost similar; in the wireless-LAN environment. In spite of this, SIP was seen to incur higher jitter values in virtually all the cases.

Table 6: comparing packet loss rates for LAN and WAN environments

cases Packet loss rate (WAN) Packet loss rate (LAN)
Non-Ideal SIP 37.4 37.4
Non-Ideal H.323 39.0 38.4
Non-Ideal Secure (PPTP) SIP 48.4 51.2
Non-Ideal Secure (PPTP) H.323 47.4 50.9
Non-Ideal Secure (IPSec) SIP 50.4 51.4
Non-Ideal Secure (IPSec) H.323 51.5 51.7

According to results shown by Markopoulou et al (2003), there is a correlation of the packet loss percentages of this research with low MOS score of about 2-3, meaning that the user will perceive a low quality. Furthermore, a comparative study by Miloucheva et al (2004) showed almost equivalent results for both SIP and H. 323 in non-ideal scenarios. According to this study, there was a significant increase of about 10 percent in terms of total sent packets, in nonideal secure scenarios.


It has been the observation of this study that QoS of VoIP communication generally decreases once network congestion is introduced, as a result of the introduced packet losses. In addition, a tangible degradation of VoIP QoS comes about due to VPN protocols. All the same, the best QoS parameter values were obtained in the network that had no congestion. However, a deterioration rate was observed in a non-ideal environment where a queuing discipline was introduced.

All the same, results obtained in a non-ideal environment are very practical and logical, given that no deployed VoIP network would ideally operate in an environment free of congestion.

In conclusion, the results obtained illustrates that dissimilar jitters are produced by call signaling protocols such as SIP and H. 323. With respect to packet losses and delays, the two signaling protocol are however significantly apart. QoS of VoIP is significantly affected by the implementation of VPN protocols; with all the three QoS parameters explored showing a deterioration of RTP packet transmission. Finally, it is worth noting here that the performance of VoIP in both LAN and WAN environments was not without the experience of jitter and delay values, especially with the implementation of VPN. Future analysis into similar work would be best placed using other optimization methods, as provided by Gardner (2003). In addition, it would also be worthwhile to consider implementing VPN over VoIP communications.


Amir, Y., Danilov, C., Hedqvist, D., and Terzis, A., “1-800 OVERLAYS: Using Overlay Networks to Improve VoIP Quality”, Technical Report CNDS-2004-2.

Bearden, M., Denby, L., Karaçah, B., Meloche, J., and Stott, T., “Experiences with Evaluating Network QoS for IP Telephony”, 10th International Workshop on Quality of Service (IWQoS 2002), USA.

Fidler, M., “On the Impacts of Traffic Shaping on End-to-End Delay Bounds in Aggregate Scheduling Networks”, Springer, LNCS 2811, Proceedings of COST 263 QoFIS, pp. 1-10, 2003.

Goode, B. “Voice Over Internet Protocol (VoIP)”. Proceedings of The IEEE. Volume 90. No. 9. 2002.

Han, J.S., Ahn, S.J., and Chung, J.W. “Study of Delay Patterns of Weighted Voice Traffic of End-to-End Users on the VoIP Network”, Int. J. Network Management 2002.

Hole, D.P. and Tobagi F.A., Capacity of an IEEE802.11b Wireless LAN Supporting VoIP, Proceedings of the IEEE International Conference on Communications (ICC) 2004.

Jha, S., and Hassan, M. “Engineering Internet QoS”. Artech House, 2004.

Karam, M.J. and Tobagi, F.A. “Analysis of the Delay and Jitter of Voice Traffic over the Internet”. Proceedings of Infocom 2001.

Kos, A., Klepec, B., and Tomažič, S. “Techniques for Performance Improvement of VoIP Applications”. IEEE MELECON 2002. Cairo, Egypt.

Miloucheva, I., Nassri, A., and Anzaloni, A., “Automated Analysis of Network QoS Parameters for Voice over IP Applications”, D41 – 2nd Inter-Domain Performance and Simulation Workshop (IPS 2004).