All existing radar systems suffer from effects of nonideality, like noise, distortions, and component or channel mismatch. These are often a result of limitations in the current level of technology, but can also stem from the desire to develop systems with the least complexity and cost. Among the most impactful effects are time and phase misalignments between individual channels in multi-antenna sensors, as well as distortions in the analog domain.

To begin with, this work provides deep theoretical knowledge about frequency-modulated continuous-wave (FMCW) multiple-input multiple-output (MIMO) radar and describes the design and construction of a prototype system that serves as a development and experimenting platform. Alongside it, a new and efficient radar simulator was developed and used for rapidly iterating over possible system architectures and testing advanced processing strategies. The radar system was built from the ground up, beginning at the base component level. Analog and digital hardware, antennas, and a capable radar software framework were created, ensuring a successful operation of the complete system. Various obstacles during the design and implementation efforts had to be overcome and are extensively discussed in this work. As a remedy to unwanted system nonidealities, novel algorithms and architectures are introduced, which improve upon the state of the art of radar signal processing. The final radar prototype was tested to achieve a maximum range of 3 km for a target of 50 m length and an average angular resolution in the order of 2.4°. With an update rate of at least 10 Hz for the complete radar image, the radar offers real-time capabilities for use in a maritime environment. This allows tracking of fast-moving targets and grants the radar operator a natural view on the sensor’s surroundings.

One of the major challenges in designing a distributed radar system with multiple transmit and receive elements, is to align the separate channels with regard to timing and phase. Differences in path length due to cables, or traces on circuit boards can lead to variations in signal propagation. Without countering these timing issues, the produced radar data is often unusable, since angular target information cannot be recovered. The proposed automatic calibration approach solves this problem elegantly, by correcting the time and phase relations of the individual channels in MIMO radar systems without the need for a physical calibration target. For achieving this, an optimizer maximizes a contrast metric that judges the sharpness of the radar image. Since the calibration quality of the radar sensor correlates with the sharpness of the radar image, it is possible to find the ideal phase correction coefficients for the individual signal paths, regardless of the target situation in front of the radar sensor.

Many radar systems have in common that they employ a number of analog components, which can alter the signal that passes through them in shape. Mixers and amplifiers, particularly in the baseband signal path of FMCW radar, introduce harmonic distortions. The presented frequency hopping approach thus tackles this impactful kind of system nonideality by cancelling harmonic contributions within the baseband, regardless of their origin. In consequence, the quality of radar images can be significantly improved without the need for specialized hardware blocks. By performing a number of only four repeated measurements with unique chirp parameters, the algorithm is able to achieve an improvement of spurious-free dynamic range of 45 dB, given a signal-to-noise-ratio (SNR) of 60 dB between the desired target beat signal and the noise floor of the frontend.