Quality of service assessment and analysis of wireless multimedia networks.

Recent years have witnessed a vast technological progress in the area of Quality of Service (QoS), mainly due to the emergence of multimedia networking and computing. QoS measurement and analysis have long been of interest to the networking research community. The major goals of this thesis are of two fold: Firstly, to investigate the effect of the QoS parameters on the overall QoS experienced by wireless networks. Secondly, to utilise the results in developing efficient mechanisms for intrusive and non-intrusive assessments of the performance of wireless ad hoc networks as well as the measurement of the available QoS for audio and videoconferencing applications over the IEEE 802.11 standard. To evaluate the network performance and the overall QoS of multimedia applications, new fuzzy logic and distance measure assessment approaches were developed taking into account the QoS parameters requirements of each application. The developed approaches essentially include measuring the main QoS parameters (delay, jitter and packet loss) and use them as input to the measurement systems, which combine them and produce an output that represents the instantaneous QoS. The devised approaches showed how the QoS can be measured without a need for complicated analytical mathematical models.In this study, several techniques were devised for estimating QoS. Firstly, a probe-based assessment method (active technique) was developed. In this method, special artificial monitoring packets were injected into the network. The overall QoS and its parameters were estimated by collecting statistics from these packets. It was possible to make reasonable inferences about the delay, throughput, packet losses and the overall average QoS using different probe rates. This technique showed some limitations for measuring the jitter. In addition, the rate of the monitoring packets played an essential role in the precision, level of resolution of estimated results and negatively impacted the network performance. Secondly, to overcome some of the drawbacks of the probing-based method, a new assessment technique was, subsequently, devised based on passive monitoring standard sampling methods. Unlike the active technique, the new method has the advantage of not adding an extra load to the network. In addition, it is not like the typical passive methods, which require the transfer and calculations of the whole captured data. Generally, all sampling schemes provided satisfactory measures of the overall QoS and its parameters and produced very acceptable bias and Relative Standard Error (RSE) result. Systematic sampling provided the most accurate estimates compared to the stratified and random approaches. In addition, after sample fraction of 2%, the estimated overall QoS bias from the actual QoS became constant and equal to -0.5% and RSE was less than 0.005 using both fuzzy and distance assessment systems. Thirdly, in order to overcome some negative aspects of inaccuracy and biasness caused by sampling techniques, a new scheme was proposed to correct these results to be closer to the actual traffic measurements. The new approach does not disturb the network performance (as in active methods), neither depends on the whole traffic (as in passive methods), nor bias the actual results (as in the standard sampling technique). Similarly, systematic sampling showed the best performance. Sample fractions, using the systematic sampling, greater than 2% gave an overall estimated QoS identical to the actual QoS because the obtained relative error was nearly constant and approximately close to zero using both assessment systems. The measured QoS can be used to optimise the received quality of the multimedia services along with the changing network conditions and to manage the utilisation of the network available resources especially for ad hoc networks. Overall, the findings of this study contribute to a method for drawing a realistic picture of the wireless multimedia networks QoS and provide a firm basis and useful insights on how to effectively design future QoS solutions.