Subject

Transient radar signal processing is critical for numerous applications in non-destructive and non-invasive testing in various industrial sectors. This master thesis focuses on leveraging advanced machine learning and deep learning techniques to enhance the quality of these unique radar signals. Specifically, the research aims to improve radar signal processing by decluttering signals and removing noise, thus providing more precise and accurate data. This will be achieved using optimization methods such as robust principal component analysis (RPCA) and deep neural networks (DNNs). The goal is to develop methodologies that can significantly enhance the interpretation and utility of transient radar signals.

Kind of work

The work involved in this thesis encompasses several tasks. First, the student will conduct a comprehensive review of existing radar signal processing methods, focusing on machine learning and deep learning techniques. Following this, the student will implement optimization methods like RPCA to address noise and clutter in specific transient radar signals. Additionally, the student will design, and train deep neural networks tailored to radar signal enhancement tasks. This will include creating and preprocessing a dataset of transient radar signals, training the models, and evaluating their performance against established benchmarks. The student will also explore the integration of RPCA with DNNs to enhance signal quality further.

Framework of the Thesis

The framework of this master thesis will be organized into several structured phases. Initially, a detailed literature review will be conducted to understand the current state-of-the-art in radar signal processing and the application of machine learning techniques in this field. Next, the student will develop a methodology for signal decluttering using RPCA and implement this methodology using Matlab and/or Python. This will be followed by the design and training of deep neural networks to improve radar signal quality further. The implementation phase will involve coding, data preprocessing, and model training. Finally, the student will evaluate the performance of the developed methods, compare them with existing techniques, and document the findings in a comprehensive thesis report. The thesis will conclude with a discussion of the results, their implications, and potential directions for future research.

A. Pourkazemi, S. Tayebi, J. Stiens, Fully blind electromagnetic characterization of deep sub-wavelength (?/100) dielectric slabs with low bandwidth differential transient radar technique at 10 GHz, IEEE Transactions on Microwave Theory and Techniques. 2021. (IF: 3.413, Q1). 10.1109/TMTT.2021.3135356.

Number of Students

1

Expected Student Profile

The ideal candidate for this master thesis should possess a strong foundation in both (radar) signal processing and machine learning. Proficiency in Matlab and/or Python is essential, as these tools will be extensively used for implementing the optimization methods and deep learning models. The student should have a good understanding of machine learning tools and Python packages, as well as experience with deep neural networks. Additionally, a solid basis in signal processing principles is crucial for understanding and addressing the challenges associated with radar signal enhancement.