Linear Frequency Modulated Continuous Wave LFM-CW Short-Range Radar for Detecting Subsurface Water Content With Deep Learning

The study is to develop a Linear Frequency Modulated Continuous Wave LFM-CW short-range radar for detecting subsurface water content with deep learning. Implementation of signal transmission/reception, signal processing, and graphical user interface in LabView. Fabrication of antenna from the design program and water table are enclosed with a Styrofoam box and buried 1m, 3m, and 5m respectively for the experiments. The experiments also involve metal and plastic buried 1m, 3m, and 5m, respectively, for data comparison. The researcher dug a 9×3×5m hole and divided it into three sections with buried objects of different deepness. The first section has a size of 3×3×5m with 2×1 metal plate, 2×1 plastic plate, and 2×1×0.5 water table box. The object is separated by 1m in a triangular manner at 5m depth from the ground. The second section has 3×3×5m with buried object 2×1 metal plate, 2×1 plastic plate, and 2×1×0.5 water table box separated by 1m in a triangular manner with 3m depth from the ground. The last section has 3×3×5m with buried object 2×1 metal plate, 2×1 plastic plate, and 2×1×0.5 water table box separated by 1m in a triangular manner with 1m depth from the ground. The results show a trend with regards to the A-scan measurement window characterizes the different dielectric properties of the water table, metal, and plastic and able to detect objects greater than 1m using the optimized systems. The deep learning method able to prove the interpreted result from the observed A-scan. The study recommends a higher bandwidth and transmitting power hardware to increased range resolution, which will be able to detect shallower objects. Consideration of ultrawideband antenna with higher directivity and gain can also improve the system subsurface detection.