Providing Strict QoS Guaranties for Flows with Time-varying Capacity Requirements

Abstract

Many distributed applications require bandwidth provisioning to implement their functionality. A prominent example is a remote real-time monitoring service in e-health system, where a special type of emergency requests require strict guaranties regarding provided transmission rates. We consider the problem of network capacity sharing between two types of flows: standard best-effort flows and QoS-constrained flows. We derive distributed control algorithms for dynamic capacity allocation allowing to serve the QoS-constrained flows by preempting the best-effort flows. Such solution minimizes the amount of unused capacity. We also present how to estimate the capacity needed to deploy QoS-based application in a way to minimize the number of flow preemptions. The presented solution is evaluated in a simulation environment.