Advances in Millimeter-wave radar processing and imaging using Machine Learning
Leuven | More than two weeks ago
To produce an image of the environment, the samples produced by radars are usually processed by low-complexity algorithms such as FFT or optimized correlator banks. Those solutions allow fast processing but make hypotheses on the received data. Unfortunately, the real world produces more complex data, which does not always meet the hypothesis behind low-complexity algorithms. This results in image blurring, defocusing or measurement artefacts. In addition, the conventional radar processing chains add constraints on the radar design to meet the set hypotheses. Especially they require many antennas to produce a high-resolution 3D image of the environment. Hence, the low complexity of the digital part of the radar requires a complex radar analogue front-end to provide high resolution. In addition, enhancing the azimuth resolution from processing ADC signals remains challenging.
On the other hand, machine learning approaches can improve radar capabilities by providing better use of radar measurements and reducing the constraints on the radar analogue front end.
To avoid the drawbacks of conventional radar processing, this PhD research aims to propose new machine learning techniques to improve the resolution of imaging radars way beyond their conventional resolution with a reduced number of antennas.
The successful PhD candidate will be part of a large IMEC team working on the research, implementation and prototyping of future radar systems: experts in digital, analogue and mm-wave ASIC design, radar systems, radar signal processing and machine learning. This is a unique opportunity to develop innovative, multi-disciplinary technology and shape future radar systems. You will publish your research in top-level journals and conferences.
Required background: Signal processing for wireless technology (communication or radar). Knowledge in machine learning. Proficiency with Matlab or Python.
Type of work: 20% literature/theory, 60% modelling/simulation, 20% experimental
Supervisor: Hichem Sahli
Daily advisor: Marc Bauduin, Hamed Javadi
The reference code for this position is 2024-064. Mention this reference code on your application form.