Enhancing Structural Health Monitoring and Management Through Edge, Fog and Cloud Computing Architectures

Abstract

The aim of Structural Health Monitoring (SHM) is to detect, locate, and quantify structural damage through data acquisition, which not only enhances human safety and minimizes infrastructure maintenance costs but can also be performed in real-time. However, environmental factors and operational risks can limit the effectiveness of SHM techniques. Fortunately, the emergence of smart connected devices and internet connectivity has enabled remote monitoring of various structures such as buildings and bridges from any location and at any time. Despite numerous advancements in Internet of Things (IoT) technology in Structural Health Monitoring (SHM), it still struggles with latency issues when analyzing and visualizing real-time data from various structures. Current cloud-based architectures are not efficient in displaying various structure metrics, such as acceleration, humidity, and temperature. Additionally, utilizing state-of-the-art machine learning techniques for predictions is time-consuming as data must travel long distances to reach the cloud. To address these challenges, emerging technologies like edge computing and fog computing can be implemented. Edge computing aims to bring processing and storage as close as possible to the application, while fog computing complements cloud computing by handling fewer intensive analytics and processing tasks. The proposed work showcases a multilayer system architecture comprising edge, fog, and cloud computing, designed to collect, analyze, and visualize sensor data via Amazon Web Services (AWS). This approach involves connecting multiple sensors to an edge device (Raspberry Pi) to develop a standalone web monitoring interface. Furthermore, machine learning algorithms are deployed on the edge to predict the local behavior of a structure using its local data. The methodology of monitoring structures using the architecture presented in this paper exhibits great potential. The effectiveness of the current approach is demonstrated through a comparative analysis of its performance and latency. Besides the software could potentially be packaged and offered as Software as a Service (SaaS) product in the future.