Analytical modelling and evaluation of generalized processor sharing systems with heterogeneous traffic

摘要

Generalized processor sharing (GPS) has been widely studied as an important scheme for providing differentiated quality of service. Realistic traffic in multi-service networks exhibits heterogeneous properties and can be categorized into two classes, namely, long range dependent (LRD) and short range dependent (SRD) traffic. Performance modelling and analysis of GPS systems subject to heterogeneous network traffic is an open and challenging problem. This paper develops an analytical performance model for GPS systems under heterogeneous LRD and SRD traffic. To this end, we propose a cost-efficient modelling method for GPS and further develop the analytical upper and lower bounds of the queue length distributions for individual traffic flows. Numerical examples and extensive simulation results are presented to validate the accuracy and merits of the analytical model.