Adaptive strategies for improvment of quality of service in IP telephony

Abstract

The convergence of multimedia communications on IP has great QoS newlineexpectations and requirements. International institution ITU-T, 3GPP newlineand IETF have emphasized the user satisfaction for the evaluation of newlineQuality of Service (QoS) and have defined the tolerance limits from the newlineuser perspective. So we require a common measurable yardstick to newlineasses and evaluate the diverse QoS requirements for the varied range newlineof applications, so that we can develop the strategies for QoS provision newlineand control on its basis. newlineThus the common measurable parameter for QoS can be the user satisfaction newlinelevel and the measurable yardstick can be the user tolerance newlinelimit. Currently there are five basic QoS parameters defined by ITUT, newline3GPP and IETF; these are delay, delay variation, information loss, newlinethroughput and sequencing. We define a mathematical model to measure newlineQoS based on user satisfaction level and try to provide a simple newlineconversion between the values of these parameters and the user satisfaction. newlineWe define the coefficients of QoS to measure user satisfaction newlinelevel and QoS. Another important consideration is the Fairness in QoS, newlinewhich implies an equal level of user satisfaction. In our model the coefficient newlineof Fairness is used to measure the fairness and it depends on newlinethe difference in the coefficients of QoS of two application flows. As the newlineQoS is assessed and evaluated at end-to-end level, but most of the QoS newlinecontrol and provisions are at hop level. Keeping this in view our measurement newlinemodel measures the QoS and fairness at each level i.e., hop newlinei newlinelevel, domain level, end-to-end network layer level and application level. newlineWe propose a traffic classification criteria based on this measurement newlinemodel. The aim of this classification is to group the traffic having similar newlineQoS requirements so that we may achieve the maximum and fair newlineuser satisfaction level. newlineWe suggest and define a QoS control architecture at hop level based newlineon queuing delay and queue size.