Pre-emptive Flow Installation for Internet of Things Devices within Software Defined Networks

Abstract

The predicted prevalence of both Internet of Things (IoT) based devices and the concept of Software Defined Networking (SDN) as a new paradigm in networking, means that consideration is required for how they will interact. Current SDN implementations operate on the principle that on receiving an unrecognised packet, a switch will query a centralised controller for a corresponding rule. Memory limitations within current switch devices dictate that this rule can only be stored for a short period of time before being removed, thus making it likely that the relatively infrequent data samples sent from IoT devices will have a transmission interval longer than this timeout. This paper proposes a Pre-emptive Flow Installation Mechanism (PFIM) that dynamically learns the transmission intervals of periodic network flows and installs the corresponding rules within a switch, prior to the arrival of a packet. A proof-of-concept implementation shows this to have a significant effect on reducing the delay experienced by these flows.