Max-min utility fair flow management for networks with route diversity

摘要

A regional central manager is employed to set aside, for the regional (or back-bone) network that it manages, for each flow class, communications capacity resources for a specific future time horizon. In the context of such a traffic management operation, a longer temporal scale is involved in controlling the admission and distribution of flows across the network. For management scal-ability purposes, flows are aggregated into flow classes. Furthermore, we consider a network operation under which multiple simultaneously activated routes are employed, across possibly distinct segments, to distribute traffic between identified source-destination pairs. We aim to ensure that the utility assigned to each class is as high as feasibly possible while striving to raise the utility gained by all classes in a max-min fair manner. In doing so, we incorporate the communications capacity constraints that are imposed by the underlying hybrid of directional and/or multiple-access wireline and wireless communications media employed across the network system. We develop and present in this paper an optimal algorithm for solving such a traffic management problem. It yields multi-utility-based max-min fair distributions of flow rates, per each class, across the specified multitude of simultaneously activated multi-segment routes. To guarantee that admitted flows are granted their desired capacity resources (and targeted corresponding utility levels), the selection of optimal flow distributions across the network routes is combined with the use of a flow admission control scheme that serves to optimally limit the aggregate rate of flows admitted for each flow class. As illustrative examples, we demonstrate the effectiveness of our solution in comparing its performance with that obtained under the use of a traffic regulation scheme that is not overlaid with a traffic management mechanism that serves to set aside resources for the support of flow classes. We also illustrate the use of our optimal algorithm for determining the optimal placement of unmanned aerial vehicle platforms that serve to supplement a terrestrial transport segment with a space-based one.