Analysis and Optimization of Product Form Stochastic Networks

点击翻译按钮获取中文摘要

Research focuses on the development of algorithms to determine the normalizing constant in product form multiclass and multirate circuit-switched networks using Monte Carlo summation. Further, gradients will be obtained. Efficient variance reduction techniques are to be developed and application of the Monte Carlo summation method to parametric optimization and dynamic control of stochastic networks investigated. Research should produce significant new discoveries in adaptive optimization algorithms and yield queueing network analysis tools that can be applied in a diversity of disciplines.