A congestion avoidance algorithm in SDN environment

Abstract

It is expected that large number of IoT devices and IoT network will be connected to the Internet. In these environments, congestion of network links could be rapidly increased due to the bust IoT traffic especially related with video devices. To avoid the congestion, currently, End-devices control the congestion links by using congestion control scheme on TCP stack such as Tahoe, Reno, NewReno, and CUBIC, However, the increased traffic connected by the IoT devices and network would much degrade the network throughput and TCP resilience. This paper proposes the flow congestion avoidance algorithm controlled by the network in SDN(Software Defined Networking) environment. In the proposed scheme, link utilization is calculated in SDN controller and recalculated rerouting algorithm is applied to switches which would be configured by using Openflow configuration protocol. To verify the improved performance, a video stream is applied to the proposed algorithm in SDN environment. We could achieve the 11% of throughput enhancement and reduced RTT in the experimental video stream application. This result refers that we could achieve improved resilience of TCP with near-lossless by controlling network.