Evaluating the functionality of an Industrial Internet of Things system in the Fog

Abstract: The Internet is one of the greatest innovations ever created by mankind, and it is a technical trend that has moved into industries to facilitate automation, supervision and management in the form of IoT devices. These devices are designed to be extremely lightweight and operate in low-power and lossy networks, and therefore run a low duty cycle and CPU-clock frequency to reserve battery life. Fog nodes are located on site to minimize network delay and provide centralized processing to handle data from hundreds of connected devices in wireless sensor networks. This is the future of industrial automation. Our goal is to show the functionality of an industrial IoT network within the scope of Fog computing by implementing a closed-loop control system in Cooja. Performance evaluations considered network reliability in terms of packet delivery ratio and timeliness. We assume that wireless IoT devices are running RPL routing (one of the most common standard routing protocols for IoT applications). We implement a mobility controller at the Fog-server in order to collect measurements made by the Fog nodes and send commands to IoT devices. In this thesis work, we assume that the commands are related to the mobility pattern of mobile node (e.g. AGVs in industrial automation) in order to avoid collision. From the simulation results we can conclude that sampling rates and node density have a greater impact on performance compared to payload size. We cannot be sure that our results reflect what a real-world evaluation would imply as we are running an emulation software, even though it has a very realistic physical layer. We do however believe that with substantial testing and improvements to both Cooja and our implementation, an accurate representation can be accomplished and algorithms in Cooja can be moved to real-world implementations.