Adaptive Flow Admission Control in a Software-Defined Network

摘要

In this paper, we have followed the idea of exploiting the highly flexible architecture of Software-Defined Network (SDN) to provide an adaptive packet Flow Admission Control (FAC) system. Both Quality of Service (QoS) requirements of the packet flow and the current state and situation in the network are put under consideration when admission decisions are made. Furthermore, our proposed FAC monitors and dynamically adapts to the changes of network parameters such as bandwidth, delay, jitter, latency, and classification of traffics for better overall user experience. This paper finds ways to implement the proposed adaptive FAC for improved QoS using a custom built OpenFlow controller based on Floodlight in the production environment of GENI, a national testbed for innovative networking and distributed systems experiments. The experimental results indicate that the adaptive FAC delivers an improvement of up to 67% in comparison with the traditional FAC when unsatisfied user percentage is measured. Also, the adaptive FAC enhanced the average end-to-end delay and network resource (i.e. bandwidth) utilization by up to 30% and 50% respectively.