During the last ten years, millimeter-wave radars have evolved from expensive bulky systems to highly integrated and miniaturized radar-on-a-chip. This evolution is largely due to safety requirements in the automotive sector. Now that simple millimeter-wave radars are becoming commonplace, a second revolution is on its way: using a large number of radar transceivers and antennas to create so-called 2D or even 3D radar images. Pushing the numbers to hundreds of transceivers, interpretable images can be generated. The holy grail will be to have radar images approaching the angular resolution of Lidars or even cameras. Increasing the 3D radar resolutions (in range, azimuth and elevation) comes with an unacceptable cost in hardware resources and power consumption. It is therefore mandatory to drastically reduce the hardware count for a given imaging resolution level.
The purpose of this PhD research is to research hardware, signal processing and machine learning solutions to improve the performance of imaging radars way beyond its conventional performance. Possible ingredients could be among hardware distributed accelerators, image denoising and sharpening, sparse array imaging (with sparsity in the frequency and/or spatial domain), feature extraction and classification, data augmentation or transfer learning, neuromorphic processors, etc...The IMEC radar testbed at 140GHz will be used to experiment and validate the proposed solutions. The application field is very broad including automotive, robotics, security, and many more.
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, analog and mm-wave ASIC design, radar systems, radar signal processing and machine learning.
Required background: Signal processing, machine learning. Desired: radar sensors.
Type of work: 80% modeling/simulation, 10% experimental, 10% literature
Supervisor: Hichem Sahli
Daily advisor: Andre Bourdoux
The reference code for this position is 2020-099. Mention this reference code on your application form.