DECENTRALIZED ROBUST FLOW CONTROLLER DESIGN FOR DATA-COMMUNICATION NETWORKS

An important problem in the operation of data-communication networks is congestion. A tool which is used to avoid congestion is flow control. A flow controller decides on the rate of messages to be admitted to the network so that congestion can be avoided. A challenging aspect of flow controller design is the existence of time-delays in the data-flow. Furthermore, these time-delays are usually uncertain, multiple, and are time-varying. In this work we first review rate-based robust flow controller design for single-bottleneck networks. We then review previous works on the extension to multi-bottleneck networks, where decentralized controllers are used. Finally, we discuss use of overlapping decompositions on data-communication networks for robust flow controller design.