Finding Periodicity of Subflows in SDN-based IoT

With the development of the Internet of Things (IoT), the amount of IoT devices in the network is growing. According to a report by McKinsey Global Institute, there will be 1 trillion IoT devices connected to the Internet by 2025. Faced with a large variety of IoT devices, it is necessary to use software-defined networking (SDN) to solve the connection problem of IoT devices. Nevertheless, the size of the flow table of the switch in SDN is limited, and it cannot accommodate all flow entries of all passing traffic. Therefore, an effective flow entry management scheme is needed to reduce the controller processing delay and signal overhead. In general networks, a suggestion of delaying installation is proposed, which can reduce the number of flow entries. However, this will increase the controller processing delay and reduce the quality of service. Two pre-installation methods designed for IoT were proposed, PFIM and EPFIM. Both of them can reduce the controller processing delay by detecting the periodicity of traffic and pre-installing flow entries. However, they cannot detect multiple periodicities. In addition, there is still room for improving the accuracy of periodic detection and reducing the signal overhead. In this paper, we propose a subflow-based proactive flow installation mechanism. We design a new data collection method to improve the accuracy of pre-installation. We also design a new data structure to be able to detect multiple periodicities in a single flow. An algorithm is proposed to implement accurate periodic calculation. Our mechanism can fit with arbitrary periodic traffic patterns and has been proven to be highly accurate and efficient.