Network calculus and service curve scheduling in heterogeneous networks

摘要

We describe a generalised framework for analysing the delay and backlog bounds in lossless packet networks using the techniques of network calculus. We outline network calculus in terms of matrix operators acting on functions, and couple these operations with the concept of service-curve scheduling. We describe how this combined framework is used to determine delay bounds imposed on packets as they traverse example networks, and compare the results with standard FIFO and virtual clock scheduling algorithms. This comparison highlights the QoS advantages to be gained from the use of network calculus and service-curve scheduling.