Compressed Sensing Applied to Non-Ideal Microwave Measurements in Metal Enclosures

Paper in proceedings, 2019

Compressed sensing can make use of a priori information in the sense that its dictionary may be constructed for an expected object of particular size and shape, should the application deal with the detection and positioning of exactly such expected objects. Here, we test such a compressed-sensing approach on measurements that include objects that are foreign to the dictionary. It is found that the compressed-sensing approach shows reasonable performance that degrades gracefully as the measurement region is contaminated by dielectric objects that are foreign to the dictionary, where the degradation also becomes more severe as more dielectric material is present in the measurement region due to mutual interactions not accounted for by the dictionary.