Fog computing based efficient IoT scheme for the Industry 4.0

Abstract

Industry 4.0 aims to dramatically enhance the productivity of manufacturing technologies through the collection and analysis of real-time data. This combines the ubiquity of the IoT with the processing capabilities of cloud computing to generate insights that help to optimize the decision making process. The increasing demand of data and the explosion in the number of sensing devices, which might be highly constrained in terms of communication, battery and computational power, introduce new challenges that need efficient IoT-Cloud architectures. With this in mind, we extend MQTT, the de facto IoT communication protocol, using a fog computing approach that introduces a low complexity computational layer between the Cloud and IoT nodes. In this approach, the MQTT broker, which is in charge of relaying data from publishers to subscribers, is placed at the fog layer. The purpose of introducing an intermediate layer to this particular scenario is to: i) predict future data measurements through prediction techniques; ii) operate as a gateway to upper layers; and iii) provide the capability to offload computationally expensive data processing jobs from the Cloud to the Fog, minimizing additional latency and operational expenses. With this architecture, the transmissions required from IoT devices may be reduced, since the publishers would only need to update the predicted data in case of mismatching. We validate our approach with an energy consumption analysis and simulations of different Machine Learning algorithms on a real dataset, and compare it with the traditional MQTT scheme.