Geiß J (2023)
Publication Language: English
Publication Type: Thesis
Publication year: 2023
URI: https://opus4.kobv.de/opus4-fau/frontdoor/index/index/docId/22623
In the last decades, radar imaging systems have received great interest in many applications. This development is driven by their robust performance in harsh environments, where other concepts such as optical sensors fail, as well as their capability to measure both distances and velocities. Especially in the automotive field, they have become a key technology to achieve the required high safety standards for highly automated driving. The major drawback of radar sensors is their limited imaging resolution. While the radial resolution depends on the system’s bandwidth, the angular resolution depends on the total size of the available radar aperture. Therefore, most sensors are already equipped with antenna arrays on both the transmit and receiving side, yielding multiple-input multipleoutput (MIMO) systems. Nevertheless, driven by the further increasing demands on the angular resolution, the research interest now focuses among others on exploiting the car’s movement to apply synthetic aperture radar (SAR) concepts to create larger apertures. The combination of both ideas then leads to single-input multiple-output (SIMO)-SAR and MIMO-SAR systems. Antenna arrays are generally impaired by imbalances between the channels caused by phase errors, amplitude errors, as well as mutual coupling between the antennas. These errors can severely impair the radar image quality. Therefore, the arrays are calibrated prior to their operation. However, this is often done as an end-of-line calibration and may differ from the ideal calibration data after installing the sensor at its operating position. Furthermore, aging, environmental influences like temperature shifts, and physical impacts can change the channel errors over time, such that an initial calibration is usually not sufficient to maintain the peak imaging performance required in safety-critical applications. Thereby, calibration errors can become the limiting factor of the image quality, which is practically achievable, especially because advanced high-resolution reconstruction techniques generally assume ideal sensors and, thus, are often sensitive to modeling errors. Hence, frequent re-calibration is necessary for maintaining high image quality. Unfortunately, most high-quality calibration approaches require measurements to far-field targets at known angles and suppressed multipath propagation. This requires highly controlled measurement environments, such as anechoic chambers, which are not widely available. Besides this, online calibration approaches exist, which have hardly any requirements on the environment but yield only poor calibration results. Consequently, there exists a lack of high-quality calibration concepts which can be performed in arbitrary environments and do not have high requirements on a reference system. Although the limitations of currently established calibration concepts are well known, the impact of non-ideal calibration parameters on the radar image is mostly unknown. Therefore, in this thesis, first, the impact of calibration errors on radar imaging systems is analyzed thoroughly. In radar imaging concepts, which shift antenna arrays to collect spatial information, calibration errors cause periodic error patterns, leading to a deterministic formation of ghost targets in the radar image. While this effect was already observed in literature, it was never analyzed in detail. This work, for the first time, fully describes the deterministic relation and verifies the findings with measurements. It is shown, that in MIMO, SIMO-SAR, and MIMO-SAR systems, under specific conditions, the calibration error power accumulates at few, predictable positions within the radar image. The number and position of these ghost targets can be predicted without requiring knowledge about the calibration error itself. Furthermore, a set of worst-case estimates are provided, to assess the impact of different error types on the image quality. The derived theoretical principles will enable future research in the fields of calibration quality monitoring, ghost target classification, as well as both online and offline calibration. The acquired knowledge about the separation of signal and calibration error power within SAR-systems using antenna arrays allows to derive a novel calibration concept in this work. The approach is based on a SAR measurement to a sparse target scene located in the near-field of the aperture. While the number of targets is assumed known, information about their position is not required and, hence, no elaborated reference system is needed. Furthermore, no anechoic environment is required, such that the calibration can be performed in almost arbitrary environments. The approach is derived for the calibration of both SIMO and MIMO systems and its capabilities are demonstrated by measurements yielding high-quality calibration results. It, thereby, presents a possible solution to the lack of low-effort calibration concepts for frequent high-quality calibration.
APA:
Geiß, J. (2023). Distortion Analysis and Calibration of Radar Imaging Systems with Non-Ideal Antenna Arrays (Dissertation).
MLA:
Geiß, Johanna. Distortion Analysis and Calibration of Radar Imaging Systems with Non-Ideal Antenna Arrays. Dissertation, 2023.
BibTeX: Download