Stochastic Networks and Parallel Simulations

这项研究的主要目标是开发制造和通信网络性能评估的模型。重点将放在对具有依赖节点、拥塞依赖路由和单元并发移动的“智能”网络进行建模上。这样的网络通常具有单元路由以避免拥塞、由于物理容量或工作负载平衡而产生的路由约束、每当下游节点拥塞时动态增加或降低节点处的处理速率、几个节点之间的并行处理、单元的拆分和合并、单元的并发移动或批处理、以及共享资源(工具、计算机文件)以进行处理。研究人员打算开发一大类包含依赖关系的部分平衡网络过程，这是现有的Jackson理论和准可逆过程无法处理的，因为它们只处理独立的节点和路线。这项工作包括刻画部分平衡网络过程的平衡行为，并利用制造和通信中出现的特定网络拓扑、节点依赖和流量控制来开发这些过程的子类。

The major objective of this research is to develop models for the performance evaluation of manufacturing and communications networks. The focus will be on modeling "intelligent" networks with dependent nodes, congestion-dependent routing, and concurrent movements of units. Such networks typically have routing of units to avoid congestion, routing constraints due to physical capacities or workload balancing, dynamically increasing or decreasing the processing rate at a node whenever downstream nodes are starved congested, parallel processing among several nodes, splitting and merging of units, concurrent movement or batch processing of units, and sharing of resources (tools, computer files) to do processing. The researchers intend to develop a large class of "partially balanced" network processes that incorporate the dependencies, which cannot be handled by the existing theory of Jackson and quasi-reversible processes, as they deal only with independent nodes and routings. The work involves characterizing the equilibrium behavior of partially balanced network processes and developing subclasses of these processes with specific network topologies, node dependencies and flow controls that arise in manufacturing and communications.