Control of High-Speed Networks

这是一个关于异步传输模式控制的项目 (ATM)网络和光网络。 我们专注于三组 高速网络运行的核心问题： ATM网络的业务整形 用于光网络的复用 具有延迟信息的反馈控制 我们将使用快速和大型网络的极限结果， 分析这三个问题。 具体来说，我们将使用 流体和扩散限制以及大的偏差界限。 交通 我们将使用一种新的策略（周期性平均速率）。 我们认为，这一战略可以大大减少 网络内部所需的排队，这种策略可以 简化网络的设计和控制。 复用 我们将研究的光网络的方法是统计的 先行复用。 这种多路复用方法的目的是 用于星星光网络。 它使用 这些网络，但不需要可调谐激光器或滤波器。 的 信息时滞反馈控制的研究目标 是确定反馈控制策略， 高速网络的操作，并量化增益 通过这些战略。

This is a project on the control of asynchronous transfer mode (ATM) networks and of optical networks. We focus on three sets of issues central to the operations of high speed networks: Traffic shaping for ATM networks Multiplexing for optical networks Feedback control with delayed information We will use the limit results for fast and large networks to analyze these three sets of issues. Specifically, we will use fluid and diffusion limits and large deviation bounds. For traffic shaping we will use a new strategy (periodic averaging of rate). We believe this strategy can reduce significantly the amount of queuing required inside the network, and that this strategy can simplify both the design and control of networks. the multiplexing approach that we will study for optical networks is statistical look-ahead multiplexing. This multiplexing approach is designed for star optical networks. It uses the connectivity properties of these networks but does not require tunable lasers or filters. The objectives in the study of feedback control with information delay are to identify the feedback control strategies that can improve the operations of high speed networks and quantify the gain achieved by these strategies.