The Road to an Open and Secure Data Exchange Infrastructure for SHM

Abstract

Today, structural health monitoring systems consist of various hardware and software components that are not always well synchronized. The paper emphasizes the need and challenges to synchronize multi time-domain and multi physical-phenomena data acquisition systems. On the way from a raw data sensor system to an independent decision-making edge device, there are still some challenges to be solved along the way. Identifying and combining technologies from foreign domains is the focus of the journey. Open and standardized communication protocols like OPC UA or MQTT are already wellestablished technologies with a high level of security. These communication protocols allow users to discover and receive structured data from compatible sensors and edge devices in the network. Moving the data processing and decision-making further to edge leads to reduced data transfer volume, lower latencies and improved privacy, security, and cost-effectiveness. This requires an extensible platform to embed custom feature extraction algorithms and to deploy trained machine learning models. ONNX, the open standard for machine learning interoperability, is a further puzzle piece on this journey. The paper shows the status of work, ongoing developments in the open standard community like OpenAE and platform strategies for advanced analytics and AI.