Abstract

The designing of broadband multimedia wireless network systems should aim at achieving maximum utilisation of wireless resources through statistical multiplexing, while, at the same time satisfying the Quality of Service(QoS) requirements of multimedia traffic. In this research, we consider a priority based scheduling strategy, suitable to the terrestrial/satellite wireless environment. The multimedia traffic is categorised into real-time (voice and video connections) and non-real-time (data connections) depending on whether it is delay sensitive or loss sensitive. The fixed size packets generated by each of the aggregated voice, video and data sources from all user terminals in an uplink beam, are modeled as a 2-state Markov modulated Poisson Process (MMPP). Using the counting process of real-time traffic, the real-time packet loss probability has been evaluated at the uplink. Based on the equation governing the non-real-time packet queueing process at the epochs of the beginning of each frame and by using an embedded Markov chain analysis, the elements of the transition probability matrix are derived. Using the matrix-geometric technique, the occupancy distribution non-real-time packet queue is evaluated. The illustrative results for different cases of traffic mix are presented. Further, we outline the analytical derivation for obtaining covariance function of number of real-time and non-real-time arrivals to a particular downstream link through the switch. We match the covariance function values at different lags with the covariance function of 2-state MMPP at corresponding lags in order to obtain the parameters of approximating 2-state MMPPs. Based on this, and using the single queue model of the uplink, we describe the procedure for evaluating the performance at the downlink. A simulation model has also been developed in order to assess the effects of the various approximations required for the analytical model.