engleski

Analysis of some Quality of Service parameters' impact on end-to-end delay measurements in VoIP network

Internet users and Internet service providers are using all kinds of tools for monitoring the communication network. Most of the measurements are done by inserting simulated traffic into the network and observing network performances. When used to measure some time-sensitive applications, like Voice over IP (VoIP), these tools in coherence with some network configuration techniques can result with inaccurate measurement result. We discuss some of those configuration techniques and their influence on delay measurements, and introduce possible solutions. To get closer to the solution of the transparent transmigration process, Internet has to accomplish the same level of quality of the transported voice, which includes minimizing the overall delay and maximizing reliability requirements on voice communication. Because of these strict requirements, it is mandatory to treat voice traffic differently than other network traffic. This can be accomplished by applying Quality of Service (QoS) techniques. QoS ensures that VoIP packets receive the preferential treatment they require The accuracy of end-to-end measurements is of great importance, because deployment of real-time services such as VoIP necessitates the knowledge of whether the quality requirements are met in the network. Some QoS techniques can endanger the accuracy of measurements. In this paper, we will present adequate procedures for reducing the impact of queueing QoS techniques on the end-to-end delay measurements. We tested the impact of some simple (e.g. priority queuing) and some more complex techniques (e.g. low latency queueing &#8211 ; ; ; LLQ) on the active measuring methods, such as ping. When using active tools for measuring network performances the measurements could potentially be harmed by network elements (e.g. routers) by giving measurement traffic a different priority treatment than observed traffic. In the test examples we can clearly see the effect of the mistreatment of the measurement traffic. measurements showed that when measuring network performances for special types of QoS-based services the measuring packet probes must be treated from the router the same way as the traffic of those services. In every other case the results will be highly inaccurate.

QoS; VoIP; delay

nije evidentirano