QoS-Aware Dynamic Flow Management in Software-Defined Data Center Networks

Abstract

AbstractIn this work, the problem of unbalanced data traffic in the presence of heterogeneous Internet of things (IoT) applications in software-defined data center networks (DCNs) is studied. In the existing literature, the presence of heterogeneous flows and switches and mobile IoT devices in software-defined DCN are not considered. Due to heterogeneity, it becomes an NP-hard problem. To address these issues in polynomial time, we proposed a data traffic management scheme, named FASCES, using the single-leader-multiple-followers Stackelberg game. In FASCES, each controller acts as the leader and the IoT applications act as the followers. Each leader and follower aim to achieve an optimal distribution of flow rules and optimal datarate, respectively. We also evaluate the existence of at least one Stackelberg equilibrium solution in FASCES. Furthermore, we evaluated the performance of FASCES while comparing it with the existing schemes through simulation. We observe that FASCES ensures a 16.67–19.45% increase in network throughput and a 4.34–9.43% reduction in network delay. Additionally, using FASCES, the per-flow delay reduces by 27.78–36.67% while ensuring a 15.37–26.91% increase in the per-flow throughput.KeywordsControllerData flowFlow managementGame theoryHeterogeneityQuality of serviceInternet of ThingsLoad balancingSoftware-defined networksData center networksHeterogeneous flowHeterogeneous switchesResource managementStackelberg game