Evolving Industrial Networks : Data-Driven Network Traffic Modelling and Monitoring

Abstract: The concept of Industrial IoT encompasses the joint applicability of operation and information technologies to expand the efficiency expectation of automation to green and flexible processes with innovative products and services. Future industrial networks need to accommodate, manage and guarantee the performance of converged traffic from different technologies. The network infrastructures are transforming to enable data availability for advanced applications and enhance flexibility. Nonetheless, the pace of IT–OT networks development has been slow despite their considered benefits in optimising performance and enhancing information flows. The hindering factors vary from general challenges in performance management of the diverse traffic for greenfield configuration to the lack of outlines for evolving from brownfield installations without interrupting the operation of ongoing processes. One tangible gap is the lack of insight into the brownfield installation in operation. This dissertation explores the possible evolutionary steps from brownfield installations to future industrial networks.The goal is to ensure the uninterrupted performance of brownfield installations on the path of evolving to the envisioned smart factories. It addresses the gap between the state of the art and state of practice, and the technical prerequisites of the integrated traffic classes for the development of an IIoT monitoring mechanism. A novel lightweight learning algorithm at the sensor level for an IIoT compliance monitoring system, together with a case study of traffic collected from a brownfield installation, provides the baseline of comparative analysis between the common assumptions and the state of practice. The identified gaps and challenges to address them directs the research for proposing a two-step aggregated traffic modelling by introducing new measurement method and performance indicator parameters for capturing the communication dynamics. Lastly, the sensor-level learning algorithm is refined with the knowledge gained from practice and research contributions to propose an in-band telemetry mechanism for monitoring aggregated network traffic.