A Distributed Flow Control with Backward Propagation: Algorithm and Preliminary Performance Evaluation

Abstract

Control of a large-scale network using a centralized approach is essentially difficult due to its large end-to-end delay, high heterogeneity of a large number of network components, low availability and/or reliability caused by network component failures. In this paper, we aim at realizing a control mechanism for both per-flow path selection and available bandwidth allocation using an autonomous and distributed approach. Per-flow path selection is to select multiple paths from the source node to the sink node such that the traffic demand by the source node can be transferred to the sink node as well as the total network cost can be minimized. Available bandwidth allocation is to decide the amount of bandwidth assigned to every link in the paths from the source node to the sink node, which are chosen by the perflow path selection. In this paper, we propose a distributed and scalable flow control mechanism called DFC-BP (Distributed Flow Control with Backward Propagation), which simultaneously solves per-flow path selection and available bandwidth allocation. We also investigate the effectiveness of DFC-BP in terms of efficiency and transient performance through simulation experiments.