The CORAL framework expands the Software-Defined Networks (SDNs) concept to the Internet of Things (IoT). This integration enables efficient end-to-end wireless communication based on dynamically optimized routing for heterogeneous mobility-aware networks. CORAL brings elastic network adaptations, such as SDN-based network discovery, topology maintenance and routing, over IoT devices to improve performance, reduce cost and resource utilization; a novel visualization platform based on Node-RED is used for illustration. Exploiting the WiSHFUL and eWINE infrastructures is double beneficial for our framework: CORAL is based on a real test-bed with innovative facilities, whilst it is enhanced with intelligence to support our controller decisions.
The CORAL framework enables elastic network operation through offloading complexity from the network protocols to the control software deployed at the surrounding fixed infrastructure, i.e., our CORAL controller. Such a facility tunes efficiently the trade-off between elasticity and complexity, since: (i) we take into account fundamental characteristics of wireless networks with mobile devices, such as signal issues and intermittent connectivity; and (ii) we support flexible and dynamic configuration of IoT devices to improve performance, resource-allocation and reduce cost. The goal is to provide dynamic end-to-end topology control and routing services in heterogeneous environments using centralized network control that exploits the global picture of the network.
Our framework, as part of the 2nd WiSHFUL Competitive Call for Experiments, has been integrated with the WiSHFUL platform and is capable of conducting experiments using the w-iLab.2 test-bed. In the proposed demo, we enhance the intelligence of our CORAL controller by exploiting the appropriate eWINE software component, i.e., the Feature Extractor for Link Quality Estimation. In a nutshell, our CORAL controller collects physical layer data from the test-bed (e.g., the wireless link parameters RSSI and LQI ), feeding the decision tree classifier, developed in the context of the eWINE project, and getting back estimations about the quality of the links. Based on the latter, the decision making subsystem specifies rules and thresholds to define efficient network paths through dynamic and flexible tuning of protocols’ parameters. At the same time, CORAL network modeler maintains an abstract view of the network based on a variety of data collected from the test-bed. The Fig. 1 depicts the CORAL architecture and the interfaces between the different components.
In details, southbound, the CORAL Controller interacts with the WiSHFUL platform that provides the required software and hardware experimentation capabilities, such as the appropriate radio- and network-control abstractions over heterogeneous wireless environments (i.e., the Unified Programming Interfaces - UPIs). Westbound, our controller exploits the feature extractor eWINE component for Link Quality Estimation (LQI) to advance its intelligence. The API interface uses server/client Sockets and JSON messages. Communication takes place in two phases: the training and operational one. Finally, northbound, the CORAL controller is connected with our dashboard, which constitutes a highly flexible GUI visualization tool based on Node-RED. The interface visualizes the WSN topology and measurements provided by the controller, while it also offers management functionalities and parameters to the user, such as real-time configuration. Our dashboard is based on the eWINE Generalized drag and drop (in Node-RED) component which improves the system’s functionality and flexibility and allows us almost instant addition of new network parameters and UPIs.
Detailed information about the code and the functionality can be found at: https://github.com/SWNRG/wishful-coral