Frequency-modulated continuous-wave (FMCW) radar is attractive for range sensing and thickness metrology, but limited sweep bandwidth constrains range resolution. Photomixing two continuous-wave lasers on a photoconductive antenna generates an intermediate-frequency (IF) beat signal whose center frequency equals their optical frequency offset and whose sweep follows the applied laser modulation. We introduce a target-free chirp-linearization method that routes a copy of the transmit signal through an fiber Mach-Zehnder interferometer with 8.9ns delay, producing a low-IF monitor that encodes the instantaneous chirp frequency. This delay-encoded readout enables per-ramp chirp estimation and digital linearization while keeping the required IF bandwidth small. We validate the approach for fixed-stand-off thickness sensing using paper samples. With 1ms triangular ramps centered at 140 GHz and sweeping 35 GHz, we achieve a root-mean-square error of 33.18 μm over 7,000 ramps. The architecture, based on direct laser modulation without external RF, is promising for future compact chip-scale photonic implementations.

Recent advancements in millimeter-wave (mmWave) radar technology have made it possible for radars to generate image-like observations, a capability that was previously challenging to achieve. These radar-generated images can be 4-D (azimuth and elevation angles + range + Doppler) if the sensor offers sufficiently fine spatial resolution. Angular resolution is the system’s ability to distinguish between objects that are in close angular proximity, while range discrimination is achieved through the large bandwidth supported by mmWave frequencies. By employing a substantial number of antenna elements, mmWave radars can perform beamforming in both transmit and receive directions, reducing interference and significantly improving spatial resolution. This allows the sensor to capture detailed information about an object’s distance, velocity, and 2-D angular position. Such capabilities are critical for applications like autonomous driving (AD), in-car monitoring, industrial automation, medical imaging, and security surveillance. This article explores emerging 4-D multiple-input–multiple-output (MIMO) imaging radar technology, highlighting its potential for cost-efficient, high-resolution image-like radar observations. It examines available sensors, their design methodologies, sensing robustness, and gaps in the current application of MIMO radar. In addition, the realization of massive MIMO radars through sparse and virtual array configurations is explored in this work. This study also considers potential new features and advancements in the field. Key topics include signal processing and array configuration design, both essential for optimizing radar performance. In addition, this article addresses various challenges and showcases ongoing industry efforts driving innovation in 4-D MIMO imaging radar systems.

Proper management of mutual interference plays an important role in the successful simultaneous operation of automotive frequency-modulated continuous-wave (FMCW) radar sensors at different vehicles. Compared to traditional interference handling concepts such as detect-and-mitigate or detect-and-avoid, the detect-and-exploit paradigm turns the originally interfering signals into signals of interest and uses them to obtain information about the environment. Following this idea, a method that implements such an interference exploitation strategy in terms of joint passive spectral sensing and localization of surrounding objects is elaborated and presented in this work. In summary, the method consists of a dedicated radar operational mode and a corresponding signal processing chain including pre-processing, beam steering-based signal component separation, maximum likelihood (ML)-inspired signal parameter estimation, and joint direction of arrival (DoA)-time difference of arrival (TDoA) based object localization. The unique advantage of the presented concept compared to over-the-air synchronization (OTAS)-based solutions is that it can also deal with interferers that change their ramp parameters over time. The applicability of the concept is both theoretically analyzed as well as practically demonstrated by means of measurements in an anechoic chamber, where the position of the interferer and an additional object in the surrounding can be determined with an accuracy of a few centimeters.

This paper demonstrates a DVB-T2 based passive radar system designed for real-time vehicle localization and traffic flow monitoring. Utilizing an existing DVB-T2 transmitter as an emitter of opportunity offers a cost-effective and non-intrusive solution for traffic monitoring applications. Incorporating road layout constraints enables precise vehicle position and velocity estimation based on measured bistatic target parameters. An experimental setup and a subsequent measurement campaign validate the system's performance in detecting and localizing multiple vehicles. Furthermore, the transition to an embedded computing platform demonstrates that real-time processing requirements are met for field deployment.

Aerial Synthetic Aperture Radar (SAR) systems are increasingly used in applications such as autonomous navigation of Unmanned Aerial Vehicles (UAVs), where accurate Above-Ground-Level (AGL) altitude information is critical for both image formation and safe flight operations. Small UAVs often lack payload capacity or power budget for dedicated altimetry sensors, motivating alternative solutions leveraging existing SAR data. This paper proposes a novel method to estimate AGL altitude directly from SAR range-profile data, without any additional hardware. The approach exploits the characteristic energy step in the range profiles caused by the nadir ground echo to infer the slant range to ground using a robust edge detector. The method is conceptually simple, computationally efficient, and readily integrable into existing SAR processing chains. Experimental validation on real airborne X-band SAR data from a fixed-wing aircraft shows that the proposed approach achieves meter-level accuracy, with a root-mean-square AGL error below 1% of the nominal flight altitude and median absolute errors on the order of a single range-resolution cell, demonstrating its suitability for terrain-relative navigation and support of autonomous flight.

Since frequency-modulated continuous-wave (FMCW) radars are widely used in automotive applications, mutual interference has become a critical challenge in dense traffic scenarios. Increasing radar density degrades target detection performance and limits the effectiveness of existing mitigation strategies. This paper proposes BanditFMCW, a learning-based smart spectrum access strategy that detects interference in real time and adaptively reallocates the radar operating band using multi-arm bandit (MAB) algorithms. In particular, a sliding-window upper confidence bound (SW-UCB) algorithm with side observations is developed to cope with non-stationary interference conditions. BanditFMCW is evaluated through Monte Carlo simulations and realistic highway traffic simulations. It is compared with the random orthogonalization (RO) bandit algorithm and five state-of-the-art mitigation strategies using four figures of merit (FOMs). The results show that SW-UCB with side observations reaches orthogonal transmit signals faster than RO while reducing both the likelihood of operating in interfered bands and increasing the expected signal-to-interference ratio (SIR) in local and global deployment settings.

Edge detection is a fundamental building block in synthetic aperture radar (SAR) image analysis. Speckle phenomena present in SAR images poses a challenge for detection and localization of edges. Hence, there is a need for robust edge detection methods specifically designed for such images. In this paper, three different models for edges present in SAR images are introduced and a new robust SAR edge detection (ROSED) method is proposed. The method uses a ratio of medians (ROM) kernel for detecting clutter boundaries combined with a modified constant false alarm rate (CFAR) target detector for point targets and line-like edges. Finally, the log likelihood ratio test is applied for precise edge localization. The results show that the proposed method is robust to biasing effects and provides more complete and accurate edges compared to other methods.

Orthogonal frequency-division multiplexing (OFDM) is a promising waveform candidate for future joint sensing and communication systems. It is well known that the OFDM waveform is vulnerable to in-phase and quadrature-phase (IQ) imbalance, which increases the noise floor in a range-Doppler map (RDM). A state-of-the-art method for robustifying the OFDM waveform against IQ imbalance avoids an increased noise floor, but it generates additional ghost objects in the RDM [1]. A consequence of these additional ghost objects is a reduction of the maximum unambiguous range. In this work, a novel OFDM-based waveform robust to IQ imbalance is proposed, which neither increases the noise floor nor reduces the maximum unambiguous range. The latter is achieved by shifting the ghost objects in the RDM to different velocities such that their range variations observed over several consecutive RDMs do not correspond to the observed velocity. This allows tracking algorithms to identify them as ghost objects and eliminate them for the follow-up processing steps. Moreover, we propose complete communication systems for both the proposed waveform as well as for the state-of-the-art waveform, including methods for channel estimation, synchronization, and data estimation that are specifically designed to deal with frequency selective IQ imbalance which occurs in wideband systems. The effectiveness of these communication systems is demonstrated by means of bit error ratio (BER) simulations.

Recounts the career and contributions of Andreas Stelzer.

In this work, a radar imaging concept exploiting interfering automotive frequency-modulated continuous-wave (FMCW) signals is presented. It employs a simple multistatic radar network consisting of two standard automotive radar nodes at an interfered vehicle. Each node passively receives and processes the interfering signals in order to form an image of the scene illuminated by the interferer. To this end, a signal processing chain consisting of both standard techniques known from passive radar imaging as well as problem-specific techniques tailored to automotive FMCW interference is proposed. While the image formation at radar node level requires only a loose time synchronization and minimal data transfer between the nodes, object detectability can be further improved by merging both images at a central sensor fusion unit. Finally, the feasibility of the concept as well as its opportunities for improved road user detection are demonstrated based on outdoor measurements.

The scope of Synthetic Aperture Radar (SAR) image registration is rapidly expanding beyond traditional multi-modal applications to include emerging domains such as SAR odometry, navigation, and SAR-based SLAM, where accurate registration between sequential SAR images is essential. In this work, we explore the feasibility of using deep neural network (DNN) featurematching models for SAR-to-SAR image registration. A new dataset of SAR image pairs was constructed to facilitate training and evaluation. Three state-of-the-art DNN models-ROMA, SuperGlue, and ELoFTR-were tested. ROMA, a dense matcher, achieved high accuracy without additional training, demonstrating strong generalization. In contrast, SuperGlue and ELoFTR performed poorly with pretrained weights but showed substantial improvement after fine-tuning on the SAR dataset. SuperGlue's rotation RMSE decreased by 35.3 % (from 0.3265° to 0.2111°), and x-translation error dropped by 55.5 % (from 6.70 m to 2.9797 m). ELoFTR exhibited even greater gains, with an 82.1 % reduction in rotation RMSE and over 95% improvement in xtranslation accuracy. All models achieved sub-meter accuracy with sub-second inference times, demonstrating the potential of fine-tuned DNN matchers for real-time SAR-SAR registration tasks.

This paper demonstrates the use of two FMCW radars operating at a center frequency of 243.6 GHz to generate tomographic images of an object under test. The FMCW radar principle is used to measure the Radon projection. Unlike the standard Radon transform, this method allows the use of either attenuation or phase as the reconstruction quantity, with phase being particularly beneficial for low-permittivity materials. The tomographic image is reconstructed using the filtered back-projection method to solve the inverse Radon transform problem. To enhance the signal-to-noise ratio and improve wave focusing, two converging lenses are employed in the measurement setup.

Sparse linear arrays (SLAs) are becoming increasingly popular due to their superior performance in achieving higher degrees of freedom (DOFs) compared to uniform linear arrays. However, most SLA design methods are inflexible for different DOFs; therefore, they are not well suited to addressing the problem of designing the desired virtual array configuration. In this article, we propose two methods for designing SLAs based on the coordinate descent algorithm with a maximum block improvement selection rule. These methods achieve the desired DOF and minimize the number of required sensors while considering the mutual coupling effects between array elements. The first method focuses on optimizing the array for minimal sensor usage, while the second method further enhances performance by reducing mutual coupling. In addition, we address the virtual array design problem in multiple-input multiple-output radar systems. The proposed methods are applied to achieve desired configurations in both 2-D planes and filled linear arrays. The results confirm that the proposed methods offer improved suppression of mutual coupling and leakage when compared to state-of-the-art approaches. Furthermore, the simulation results demonstrate that the desired virtual arrays can be reliably constructed in various scenarios.

Calibration and verification of radar systems usually requires well defined targets to calibrate and to evaluate the system capabilities. This paper compares active and passive radar reflectors and their field of application. Theoretical considerations regarding the aperture efficiency are made and compared over frequency. Wavelength and radial resolution are found to be a key factor for surrounding clutter levels. It is shown that for single-digit GHz, active reflectors have advantages compared to passive structures. In contrast, passive targets can be better suited for higher frequencies that are used in automotive bands due to technical simplicity and sufficient radar-cross-section.

Suppressing sidelobes in an antenna array reduces interference, enhances signal clarity, and improves detection. This paper addresses the challenge of multibeam sidelobe level reduction in linear arrays by adjusting its element spacing. Using the coordinate descent (CD) algorithm and Chebyshev polynomial approximation, we tackle the sidelobe level suppression optimization problem while considering the implementation spacing constraints between adjacent elements. Additionally, we confirm the solution convergence of the proposed method. Furthermore, we demonstrate that the proposed method yields promising results in terms of array geometry criteria compared to the state-of-the-art methods.

Phase shifters (PSs) are essential for ensuring accurate operation of phase-coded multiple-input multiple-output (MIMO) continuous-wave (CW) radars. Since practical PSs deviate from their ideal behavior and introduce phase errors, calibration is necessary for accurate radar performance. In this paper, we propose a methodology that exploits the usage of the intermediate frequency (IF) signal offset stemming from the leakage phenomenon to calibrate the phase of a transmitter PS on-chip. Our built-in self-calibration methodology works with both frequency-modulated continuous waves (FMCWs) and CWs, and is validated through MATLAB simulations as well as measurements on a state-of-the-art $76-81 \text{GHz}$ radar sensor.

Radiometric sensitivity describes the resolution capabilities of a radiometric system and is, therefore, one of the most important properties of a total power radiometer. This study investigates how different nonidealities influence this quality measure, both theoretically and through simulations and measurements. The novelty of this work lies in applying the theoretical framework of the Cramer-Rao lower bound to derive the sensitivity of a digital total power radiometer. Various receiver chain imperfections, such as additive receiver noise, gain variations, phase noise, and quantization noise, are incorporated into the derivation. Additionally, a suboptimal, yet practically relevant estimator is evaluated for its estimation capabilities.

In this work, we develop a method for robust single-cycle measurement velocity vector estimation for automotive radar. Building upon our previous work, we introduce a methodology that leverages spatial diversity for accurate estimation of the velocity vector of targets in the medium to close ranges. We extend our initial conceptual framework, addressing limitations from our first approach and proposing necessary enhancements for real-world applicability. Our improved process excels in target separation, identification, and velocity vector estimation, proving effective across various scenarios and minimizing errors. The system, tested on pedestrians and metal targets, presents a promising avenue for exploring its performance with varying target sizes. Simultaneously, our in-depth study on Doppler-multiplex modulation reveals new relevant constraints, prompting a modulation change for improved response separation. Despite the necessity of increasing module numbers for enhanced performance, our structured approach to target itemization and classification positions our methodology as a valuable framework for future systems, offering a comprehensive solution to diverse challenges in target estimation and classification within the automotive landscape.

This paper explores the feasibility of GPU accelerated near real-time image formation of spotlight SAR data acquired by small satellites using the backprojection algorithm. By introducing an approximation for the differential range, the task is computable in single precision. We demonstrate how this allows one to incorporate digital elevation models into the image formation process to reduce geometric distortions and enhance the geospatial accuracy of the resulting images. We provide figures for processing speed and energy consumption as tested on a Nvidia Jetson AGX Orin.

In this paper we propose a multi-session simultaneous localization and mapping algorithm using multiple input multiple output synthetic aperture radar images. Our algorithm uses only radar data to calculate odometry, loop-closure and inter-session constraints as well as to generate a map of the traversed environment. The proposed algorithm was validated through real-world data displaying better overall trajectory estimation, e.g., 79.3% improvement in terms of mean absolute error, as well as expanded radar-generated maps.