Introducing a cloud based architecture for the distributed analysis of Real-Time Ethernet traffic

The use of industrial communication protocols based on Real-Time Ethernet (RTE) standards is completely replacing traditional industrial fieldbuses. As usual, when a technology becomes mature, the need of efficient diagnostic and maintenance tools quickly raises. Very often, following the paradigm of Industry 4.0, the most effective diagnostic systems are today based on distributed, cloud-centric, architectures and artificial intelligence. However, the distributed analysis of RTE systems is challenging, considering the plurality of protocols and the stringent cost constrains which are common in industry. In this paper, a new architecture for the distributed analysis of RTE networks is proposed, leveraging on distributed probes that send traffic samples to Matlab cloud for remote analysis. The paper also proposes a software conversion tool to adapt general PCAP files captured by popular sniffers (e.g. Wireshark) into MAT file for easier Matlab elaboration. Last, a test bench for characterization (in terms of transfer delays) of the first part of the chain for RTE traffic sampling is described. The results show that in less than 10 seconds it is possible to transfer chunks of RTE traffic data (captured on industrial networks with hundreds of real-time devices) directly to the cloud and to have them converted in Matlab format.