Queueing Methods for Analysis and Scheduling of Transport Rings

这个项目将研究随机传输环，使用现代多类排队网络模型和方法来执行这些系统的分析和调度。研究的第一部分涉及对能够在系统中实现尽可能高生产率的调度策略进行分类。这样的策略被称为吞吐量最优。在第二阶段，目的是找出在平均在制品、延迟或保持成本等性能指标下接近最优的好的调度策略。在最后阶段，该计划将与位于德克萨斯州奥斯汀的研究联盟International SEMATECH合作，为300 mm晶圆厂使用的自动化材料处理系统(AMHS)建立逼真和详细的模拟模型。我们的目标是看看从排队分析中获得的见解和结果是否可以应用到更现实的系统中。这项研究有可能在几个领域产生更广泛的影响。第一个是半导体晶圆制造行业。在晶圆厂调度作业需要制定有效的调度、设置和分批策略，并解决许多其他问题。在最新一代的晶圆厂中，加工300 mm晶圆，材料处理完全自动化。为AMHS设计高效的调度规则提供洞察力是该项目的一个主要动机。更广泛影响的一个关键是与SEMATECH合作，后者为全球半导体行业的很大一部分制定标准，并提供基准测试研究。因此，如果研究的模拟阶段是成功的，与SEMATECH的合作创造了一个极好的机会来说服目前正在运营的晶圆厂在AMHS软件中实施结果。第二个动机出现在连接计算机的局域网的设计中，这也可以用本研究中提出的系统来建模。借助于排队论开发的算法已经在许多实例中在实际的互联网路由软件中实现。这一以往的成功表明，在电信领域广泛传播是相当可行的。

This project will study stochastic transport rings, using modern multiclass queueing network models and methods to perform analysis and scheduling of these systems. The first part of the research involves classifying scheduling policies which can achieve the highest possible production rate in the system. Such policies are called throughput optimal. In the second stage, the intent is to identify good scheduling policies which are near optimal under some performance measure like average WIP, delay, or holding cost. In the last stage, the plan is to work with International SEMATECH, a research consortium in Austin, Texas, to build realistic and detailed simulation models of the automated material handling system (AMHS) used in 300mm wafer fabs. The goal is to see if the insights and results obtained from the queueing analysis can carry over to more realistic systems.This research has the potential for broader impact in several areas. The first is in the semiconductor wafer fabrication industry. Scheduling jobs in wafer fabs requires formulating efficient dispatch, setup, and batching policies and addressing a host of other issues. In the newest generation of wafer fabs, processing 300mm wafers, material handling is completely automated. Providing insight into the design of efficient scheduling rules for the AMHS is one major motivation for the project. A key to broader impact is working with SEMATECH, which sets standards and provides benchmark testing of research for a large portion of the global semiconductor industry. Hence, if the simulation stage of the research is successful, collaboration with SEMATECH creates an excellent opportunity to convince currently operating wafer fabs to implement the results in AMHS software. A second motivation arises in the design of Local Area Networks linking computers, which can also be modeled by the systems proposed in this study. Algorithms developed with the aid of queueing theory have been implemented in actual internet routing software in a number of instances. This previous success indicates that broad dissemination in the telecommunications arena is quite feasible.