Performance Evaluation and Policy Selection in Multiclass Networks

This paper concerns modeling and policy synthesis for regulation of multiclass queueing networks. A 2-parameter network model is introduced to allow independent modeling of variability and mean processing-rates, while maintaining simplicity of the model. Policy synthesis is based on consideration of more tractable workload models, and then translating a policy from this abstraction to the discrete network of interest. Translation is made possible through the use of safety-stocks that maintain feasibility of workload trajectories. This is a well-known approach in the queueing theory literature, and may be viewed as a generic approach to avoid deadlock in a discrete-event dynamical system. Simulation is used to evaluate a given policy, and to tune safety-stock levels. These simulations are accelerated through a variance reduction technique that incorporates stochastic approximation to tune the variance reduction. The search for appropriate safety-stock levels is coordinated through a cutting plane algorithm. Both the policy synthesis and the simulation acceleration rely heavily on the development of approximations to the value function through fluid model considerations.