M. Sc. Jiyeon Choi

Introduction

Radar systems have become a key sensing technology in various applications such as autonomous driving, surveillance. The deep learning and machine learning techniques have shown significant potential in enhancing radar signal processing performance beyond conventional signal processing approaches.

This thesis focuses on the development and evaluation of deep learning–based algorithms for radar signal processing, with an emphasis on target detection, tracking, and classification. The research investigates how data-driven methods as well as model driven methods can effectively extract meaningful features from radar signals and improve robustness under challenging conditions such as clutter, noise, and multipath propagation.

The study may include the design of neural network architectures tailored for radar data (e.g., range-Doppler maps, point clouds, or raw I/Q signals) and a comparative analysis against conventional signal processing techniques. Depending on the scope, reinforcement learning approaches for adaptive radar processing or tracking strategies can also be explored.

Requirements

Background knowledge in radar systems and radar signal processing

Basic understanding of deep learning or reinforcement learning

Basic skills in MATLAB or Python for simulations and performance evaluations