A Software-Defined Networking framework for IoT based on 6LoWPAN Fabian Fernando Jurado Lasso, Ken Clarke, Ampalavanapillai Nirmalathas Department of Electrical and Electronic Engineering The University of Melbourne Melbourne, VIC 3010, Australia fdo.jurado@gmail.com Abstract—— The software defined networking framework facilitates flexible and reliable internet of things networks by moving the network intelligence to a centralized location while enabling low power wireless network in the edge. In this paper, we present SD-WSN6Lo, a novel software-defined wireless man- agement solution for 6LoWPAN networks that aims to reduce the management complexity in WSN’s. As an example of the technique, a simulation of controlling the power consumption of sensor nodes is presented. The results demonstrate improved energy consumption of approximately 15% on average per node compared to the baseline condition. Index Terms—Wireless sensor networks; Internet of Things; software-defined networking; 6LoWPAN. I. I NTRODUCTION The concept of the Internet of Things (IoT) is enabled by the connectivity of a variety of objects, which are embedded with processing, sensing and communication capabilities, so they can work cooperatively to accomplish a common task [1], [2]. The integration of new enabling technologies and innovative applications are forming a genuine IoT [3]. IoT applications include precision agriculture, smart cities and smart transporta- tion systems, require the deployment of thousands of sensor nodes. In fact, in 2011 the number of interconnected objects overtook the number of people [3]. However, smart objects are often constrained by processing power, memory capabilities and power consumption. Moreover, to have an IoT network with WSN devices, these devices must be accessible individ- ually by a unique Internet Protocol (IP) address. To alleviate the aforementioned issues and satisfy the IoT requirements and challenges such as scalability, heterogeneity and bridge the gap between research and practical implementation [4], the IETF (Internet Engineering Task Force) has established the 6LoWPAN [5] (IPv6 over Low-Power Wireless Personal Area Networks) working group. The 6LoWPAN adaptation layer is crucial to enable connectivity of an increasing number of objects since it allows transmission and reception of IPv6 packets over IEEE 802.15.4 based networks. In addition, due to its lightweight implementation and capability to enable interoperability between heterogeneous devices, 6LoWPAN has been adopted by many M2M communication systems [6], [7]. For such applications, there is a need for a WSN manage- ment system to ensure the network runs smoothly and is easy to maintain and manage, while also allowing accurate and efficient modification of different parameters in the network as circumstances or requirements change. This is the paradigm of Software-Defined Wireless Sensor Networks (SD-WSN). The principle of SDN is to decouple the control plane from the data plane in the network. The control plane manages complex network operations, while the data plane performs basic operations such as packet forwarding [8], [9]. Previous efforts by other groups in this domain have taken various approaches to enable reprogrammable nodes in WSNs. Luo et al. [10] proposed Sensor OpenFlow (SOF) as a com- munication protocol between the control and data planes. The data plane is made reprogrammable by customizing the flow- table of every node by using the SOF protocol, and the control plane centralizes the network intelligence in the controller. Two solutions for flow creation have also been proposed by this group. Their first solution redefines flow tables, which classifies WSN addressing by using ZigBee 16-bit network addresses and concatenated value pairs. Their other solution is to augment the WSN with IP. However, they do not provide any performance metrics evidence in the form of a simulation or practical implementation. Gallucio et al. [11] proposed SDN-WISE to reduce control information exchange between the controller and the sensor nodes. Furthermore, sensor nodes can be programmable as finite state machines which enables them to make decisions to reduce the interaction with the controller. However, this solution has not been implemented in real sensor nodes such as TelosB [12], Zolertia Z1 [13], etc. De Oliveira et al. [14] proposed TinySDN which enables multiple controllers in the WSN. It is based on TinyOS and it consists of: the SDN-enabled sensor node, which is a data plane component, and the SDN controller node, which is the control plane component where all the intelligence resides. TinySDN was designed to be hardware independent. However, they do not provide any performance metrics regarding the benefits that SDN can offer to WSN’s as the evaluation they provide lacks sufficient detail. In this paper, we demonstrate for the first time and im- plement on Contiki OS [15], a software-defined wireless sensor network framework for 6LoWPAN (SD-WSN6Lo). It consists of two main components: an SDN Sensor Node, which forwards packets based on received information from the controller, and an SDN Controller Node, where all the 978-1-5386-3395-3/18/$31.00 ©2018 IEEE