Millimeter-Wave Radar data processing

Sammanfattning: Based on the feasibility study [4], the goal of this master thesis is to compile custom firmware for the XM112 Pulsed Coherent Radar Module, which features a state of the art radar sensor integrated with a microcontroller unit (MCU). This allows the use of custom firmware to check the plausibility of real-time data processing on the radar signals and will lay some of the groundwork for further development of detection algorithms. In the feasibility study, machine learning was used in the implementation of real-time detection of wrong foot placements while walking on stairs with impressive results like less than 1 percent false positives. The feasibility study on fall prevention used pre-recorded data which was transmitted over long wires to a system of two computers linked with wifi that handled the algorithms and data processing. Being able to process data directly on the MCU will improve upon some of the redundant setup that was used on the feasibility study. The XM112 radar, which is used in this thesis, uses mm-waves which operate in the 30-300 GHz interval. The mm-wave technology works without a need for physical contact with the measured object and can provide information about the velocity, range and angle. The signals can pass through certain materials like clothing, plastic, and they are not affected by weather. Due to the short wavelengths used the antenna is also very small, resulting in a system which is very portable and can be fitted almost anywhere.[1] This thesis will implement firmware based on the provided radar system software (RSS) from the manufacturer. The firmware will contain a selection protocol using serial communication and real-time data processing directly on the XM112 Microcontroller Unit.