An Edge-Based Machine Learning-Enabled Approach in Structural Health Monitoring for Public Protection

Abstract

5G technologies have opened a wide range of possibilities in all the application fields that require features as low latency and massive Machine-Type Communications (mMTC), such as Structural Health Monitoring (SHM) systems. In this paper, an edge-based Machine Learning (ML) enabled public safety service is proposed where an SHM system is exploited to support public protection actions in case of critical situations. To this aim, an end-to-end solution based on ultra Reliable and Low Latency (uRLLC) networks is proposed together with an innovative ML-based approach that uses SHM systems information to detect critical issues in structures. The unprecedented level of reliability offered by uRLLC networks together with the efficient ML modeling capabilities allow to efficiently propagate an alarm message in case of emergency. Referring to the 5G vision, the proposed SHM system can thus be considered depending on the operational scenario: in the case of data collection and processing from sensors, considering the high number of sensors installed, it can refer to the massive Machine-Type Communications (mMTC) context; vice-versa, during a safety critical situation e.g., during an earthquake or under structural problems, or immediately after the event, the use case requires high reliability, connectivity, and sometimes low latency, i.e. uRLLC.