Mohamed Ait Gacem

Moisture ingress within fiberglass insulation is a major contributor to corrosion under insulation (CUI) in steel pipelines, leading to structural degradation, safety risks, and significant maintenance costs. Conventional inspection methods are often invasive, labor-intensive, or insufficiently sensitive to early-stage moisture accumulation. This paper presents a compact microwave sensing approach that repurposes commodity Wi-Fi transceivers operating in the 2.4 GHz band for non-destructive moisture detection in fiberglass-insulated pipelines. Instead of using the Wi-Fi microwave signal solely for data communication, the proposed method exploits the phase difference measured between two receiving antennas for a transmitted signal to detect dielectric perturbations caused by moisture ingress within the insulation layer. Both the phase-difference feature and the conventional amplitude response are extracted from the Channel State Information (CSI) of the same transmitted and received Wi-Fi packet, enabling a direct comparison between the proposed method and traditional amplitude-based microwave inspection under identical measurement conditions. System performance was evaluated through full-wave electromagnetic simulations using Ansys HFSS and validated experimentally across variations in antenna standoff distance, pipe radius, and insulation thickness. Results show that the proposed phase-difference sensing method achieves a 10% improvement in detection accuracy compared with amplitude-based microwave inspection, eliminating ambiguous measurement regions observed in amplitude responses. The method also demonstrates sensitivity to localized moisture inclusions as small as approximately 0.8 ml while maintaining clear discrimination between dry and moist insulation states. These results highlight the potential of Wi-Fi-based phase-difference microwave sensing as a low-cost, non-invasive solution for real-time moisture monitoring in industrial pipeline inspection systems.

The process of identifying the physical properties of raw construction materials is vital in several industrial and quality assurance applications. Ideally, this process needs to be performed without damaging the sample and at low-cost, while obtaining high-accurate results. In this work, a novel non-destructive construction materials classification tool is proposed. The proposed method is based on passing Wi-Fi signals through the observed samples, then analyzing the Channel State Information (CSI) magnitude and phase components. The collected CSI data packets are pre-processed by performing an averaging operation. Then, the resulting data are divided into training and validation sets and used to train Convolutional Neural Networks (CNNs). Here, the trained CNN models are formulated either as classifiers or regression models, depending on the material under test. If the objective is to sort materials within specific classes, then the CNN is formulated as a classifier. Alternatively, if the goal is to estimate a continuously varying parameter in a material, then the CNN is formulated as a regression model. Furthermore, as per the collected data, the proposed method is used to identify the construction materials based on their thickness, water content value (moisture), and compaction. The obtained experimental results effectively demonstrate the potential and merits of the proposed method. Overall, the trained CNN models achieved a 100% validation accuracy and a low validation loss, which confirms that the method is valid and highly accurate.