Traffic modeling in a multi-media environment

The evolving Broadband Integrated Service Digital Networks (B-ISDN) provide bearer service capabilities supporting real time traffic such as voice and video traffic along with jitter tolerant traffic such as data traffic. These wide spectrum of traffic sources exhibit a diverse mixture of traffic characteristics and have varied quality of service requirements. Designing and managing these evolving networks requires predictions of network performance. Traffic source models capture enough essential properties of the source, that a trace of equivalent traffic can be generated artificially. These models could then be used to investigate many of the open issues in the evolution of B-ISDN. Appropriate models that characterize the variability and correlations in the aggregate (or integrated) traffic are imperative and play an important role in the successful engineering of the future broadband networks.;In this thesis a traffic generator that can represent the behaviour of multi-media traffic for the purpose of evaluating some queueing systems, is developed. The statistical issues involved in the packet arrival process in multi-media networks are explained. These aspects are considered in the development of the traffic generator. A new traffic model called the PMPP (Pareto modulated Poisson process) is proposed. This model can capture the self-similarity and long range dependence characteristics. The traffic generator developed uses MMPP (Markov modulated Poisson process) to characterize voice and video traffic. The long range dependent data traffic is characterized by the proposed PMPP model. The developed traffic generator is used to study the queueing characteristics of long range dependent traffic and its effect on multi-media traffic.