TaDPole: Traffic-aware Discovery Protocol for Software-Defined Wireless and Mobile Networks

Abstract

The Software-Defined Networking (SDN) technologies enhance the performance, reliability, and cost of managing the functions, controls, and services of the wireless and mobile network infrastructures (i.e., Internet of Things). However, the current OpenFlow Discovery Protocol (OFDP) in SDN poses substantial scalability, accuracy, and latency challenges due to its gossipy, centralized, periodic, and tardy protocol nature. Furthermore, the problems are aggravated in the wireless, and mobile SDN due to the dynamic topology churns and the lack of link-layer discovery methods.In this paper, we design and build a novel Traffic-aware Discovery Protocol (TaDPole) for wireless and mobile SDN. We facilitate multiple discovery frequency timers for each target instead of using a uniform discovery timer for the entire network. TaDPole calculates the significance of each discovery target according to the recent network usage by assuming that the higher traffic node has more impact on the network service. It lessens discovery delay by increasing the discovery frequency to the more critical nodes. Also, it enhances the control message efficiency by reducing the discovery frequency to the less significant targets. Besides, it supports the port-neutral broadcast-based discovery method instead of using port-specific request and response approaches. We have implemented TaDPole on the RYU controller. Extensive Mininet experiment results validate that TaDPole improves discovery message efficiency by two times and makes the control traffic less bursty than OFDP with a uniform timer. It reduces the network status discovery delay by three times without increasing the control overhead.