Automotive frequency-modulated continuous wave (FMCW) radars, essential in Advanced Driver Assistance Systems, encounter mutual interference issues that degrade their detection capabilities. Model-based algorithms, though widely used, rely heavily on predetermined assumptions about the statistical properties. General-purpose black-box deep learning approaches, while effective in their training distribution, often lack flexibility and generalizability in dynamic environments. We introduce a novel hybrid method that combines model-based techniques with deep learning, treating interference mitigation as a source separation problem. Specifically, our method employs score-based deep generative networks to accurately capture the structure of FMCW interference. Additionally, we employ deep unfolding to accelerate inference, critical for automotive radar applications. Empirical results from simulated data demonstrate that the proposed algorithm outperforms the baseline models by 3.26 dB in signal-to-interference-plus-noise ratio in the presence of aggressive interference, and also shows good generalizability with measured data.

Angle estimation performance is highly sensitive to the phase and amplitude errors within the antenna array, which can arise from channel imbalances and mutual coupling between array elements. Large phase and amplitude errors can cause bad angle accuracy, high angular sidelobes, wrong estimation of number of targets, and even angle jumps, particularly in sparse array configurations. Typically, the phase and amplitude imperfections are compensated using a calibration matrix. This paper presents a null space method to calculate the calibration matrix with enhanced accuracy. The effectiveness of the null space method is verified by both simulation data and real radar data.

Radar is a key element for Advanced Driving Assistance Systems (ADAS) and Autonomous Driving (AD) systems that precisely locates objects and measures their velocities, even in inclement weather. Latest developments in the field of AI/ML have raised interests in improving automotive radar performance using data-driven algorithms, especially for direction of arrival (DoA) estimation. AI/ML methods, however, could suffer from safety-related issues, which are critical for ADAS/AD technologies. The safety concerns regarding AI/ML for DoA estimation have been addressed in a methodical and thorough manner. And an approach of combining an AI/ML model with a conventional analytic model for DoA estimation is proposed.

Interference mitigation is crucial to restore the degraded detection performance of interfered radars. For developing advanced interference mitigation methods, a well-defined evaluation methodology is required to assess the performance of different methods thoroughly and appropriately. In this paper, we propose to evaluate interference mitigation methods by measuring detection performance and signal-to-noise ratio in the range-Doppler domain. We especially suggest measuring the performance analytically without involving any detector but based on the estimated probability density functions of the target and background to isolate the effect of the detector in the performance evaluation. The time-domain thresholding and time-frequency domain thresholding methods are compared to validate the proposed methodology on simulated data.

In this paper, we introduce an enhanced version of the Bayesian linear regression with Cauchy prior (BLRC) algorithm, as previously detailed in [1]. This refinement, termed efficient BLRC (E-BLRC), is specifically designed to satisfy the stringent demands for real-time processing and hardware compatibility in automotive radar systems. Utilizing a novel "soft pruning" technique, E-BLRC effectively balances computational and memory demands with the accuracy of target angle estimations, making it a practical and effective solution for solving direction-of-arrival (DoA) estimation problem in automotive radar.

In this work, we proposed a W-band substrate integrated waveguide (SIW) antenna design with periodically shaped via wall, which improves the matching and provides additional design freedom without extra circuit area nor vias. Radar system with less transmitting/receiving channels can be realized utilizing the proposed frequency scanning SIW antenna. To validate the performance, full-wave simulation of the antenna and preliminary radar system point cloud simulation are carried out, where a good agreement can be found between the simulated point cloud and ground truth.

Automotive radar interference is a growing problem as automotive radars proliferate in advanced driver assistance systems and autonomous driving. Numerous studies have been proposed to address interference mitigation based on hand-crafted priors, like sparsity-based techniques, or through purely data-driven approaches. However, their effectiveness is often compromised when these representations fail to accurately reflect the statistical characteristics of the interfering radar parameters in dynamic scenarios. In this work, we propose a new method that treats interference mitigation as a source separation problem. We leverage score-based generative networks to explicitly learn the interfering radar parameters. These learned parameters are subsequently combined with Maximum-a-posteriori estimation, allowing for an algorithm with enhanced performance. We demonstrate that our algorithm outperforms the baselines in signal-to-noise ratio.

Compressive sensing has allowed for the improve-ment of angular resolution in radar technology, which involves two aspects: sparse signal recovery and measurement matrix design. Assuming a sparse target scene, compressive sensing radar depends solely on the design of the measurement matrix to possess certain properties, such as satisfying the restricted isometry property (RIP) and low coherence. The design of the measurement matrix depends on the location of the antennas. In this work, we consider the antenna placement problem in compressive sensing radar. The problem is interpreted as a binary program, where we propose to solve it directly using a heuristic binary optimization algorithm. The proposed binary differen-tial evolution (BDE) algorithm is able to navigate the search space with relatively high diversity while still refining promising candidates. Results illustrate the superiority of approaching the problem directly using BDE rather than resorting to relaxation approaches in the literature.

In automotive radar, time-domain thresholding (TD-TH) and time-frequency domain thresholding (TFD-TH) are crucial techniques underpinning numerous interference mitigation methods. Despite their importance, comprehensive evaluations of these methods in dense traffic scenarios with different types of interference are limited. In this study, we segment automotive radar interference into three distinct categories. Utilizing the in-house traffic scenario and automotive radar simulator, we evaluate interference mitigation methods across multiple metrics: probability of detection, signal-to-interference-plus-noise ratio, and phase error involving hundreds of targets and dozens of interfering radars. The numerical results highlight that TFD-TH is more effective than TD-TH, particularly as the density and signal correlation of interfering radars escalate.

This study introduces two novel coherent-signal single-snapshot sparse array high resolution angle finding (HRAF) algorithms: the Binary Search Matching Pursuit (BSMP) and the Bayesian Linear Regression with Cauchy Prior (BLRC). To provide a comprehensive and quantitative evaluation of these HRAF algorithms, we develop four key performance indicators (KPls): the probability of target separation, the 95th percentile direction-of-arrival (DoA) estimation error, the 95th percentile number of spurious targets, and the 95th percentile power of spurious targets. The analysis conducted in this study not only highlights the strengths and potential applications of the BSMP and BLRC algorithms but also sets a new benchmark for evaluating various HRAF algorithms in the context of advanced automotive radar systems.

Automotive radar with sparse arrays are highly desired, as a sparse array has a smaller number of elements compared to a uniform linear array (ULA) of the same physical aperture size, resulting in lower system cost, increased flexibility, and reduced mutual coupling between antennas. However, this leads to an increase in sidelobes in the angle spectrum and higher signal processing complexity. Interpolation techniques can help reduce sidelobe levels, mitigating ambiguity in angle estimation with sparse arrays, and improving the ability to distinguish desired signals from interference. In this paper, we investigate transform matrix optimization technique to interpolate a virtual sparse array (VSA) synthesized by automotive multi-input multi-output (MIMO) radar to a ULA so that high-resolution direction-of-arrival estimation algorithms that are designed for ULA can be applied to VSA to achieve high-resolution radar imaging. Our simulation results in the one dimensional (1D) sparse arrays demonstrate the feasibility and effectiveness of this technique.

Aiming at the problems of the low robustness and poor reliability of a single positioning source in complex indoor environments, a multi-level fusion indoor positioning technology considering credible evaluation is proposed. A multi-dimensional electromagnetic atlas including pseudolites (PL), Wi-Fi and a geomagnetic field is constructed, and the unsupervised learning model is used to sample in the latent space to achieve a feature-level fusion positioning. A location credibility evaluation method is designed to improve the credibility of the positioning system through a multi-dimensional data quality evaluation and heterogeneous information auxiliary constraints. Finally, a large number of experiments were carried out in the laboratory environment, and, finally, about 90% of the positioning error was better than 1 m, and the average positioning error was 0.56 m. Compared with several relatively advanced positioning methods (Inter-satellite CPDM/Epoch-CPDS/Z-KPI) at present, the average positioning accuracy is improved by about 56%, 83.5% and 82.9%, respectively, which verifies the effectiveness of the algorithm. To verify the effect of the proposed method in a practical application environment, the proposed positioning system is deployed in the 2022 Winter Olympics venues. The results show that the proposed method has a significant improvement in the positioning accuracy and continuity.

In this article, a sparse signal recovery algorithm using Bayesian linear regression with Cauchy prior (BLRC) is proposed. Utilizing an approximate expectation maximization (AEM) scheme, a systematic hyperparameter updating strategy is developed to make BLRC practical in highly dynamic scenarios. Remarkably, with a more compact latent space, BLRC not only possesses essential features of the well-known sparse Bayesian learning and iterative reweighted $l_{2}$ algorithms but also outperforms them. Using sparse array and coprime array, numerical analyses are first performed to show the superior performance of BLRC under various noise levels, array sizes, and sparsity levels. Applications of BLRC to sparse multiple-input and multiple-output radar array signal processing are then carried out to show that the proposed BLRC can efficiently produce high-resolution images of the targets.

A general non-asymptotic theoretical analysis is developed for fingerprinting localization system designs. Based on this analysis, hybrid fingerprinting and propagation-based methods are proposed using 5G-like received signal strength (RSS), time of arrival (TOA), and direction of arrival (DOA) measurements which have been a focus in recent 3GPP Rel 16 and Rel 17 positioning activities. The proposed hybrid methods have the flexibility and robustness of fingerprinting methods in dealing with none-line-of-sight (NLOS) problem while inheriting the efficiency and accuracy of propagation-based methods in 3-D localization. Specifically, a ray extension technique is developed as the propagation-based method. Then the ray extension is combined with two fingerprinting methods, the conventional weighted k-nearest neighbors (WKNN) and the proposed optimal WKNN (OWKNN), in order to remedy the geometrical deficiency in fingerprinting methods. Based on the non-asymptotic study, the proposed hybrid methods are guaranteed to outperform the fingerprinting methods without ray extension. Verification of the proposed methods is performed in a large none-line-of-sight (NLOS) urban San Jose region using simulation data provided by a previously developed super-efficient ray launcher.

Radar is imperative for many automotive applications in detecting targets. Accurate direction of arrival (DOA) estimation is essential for maximizing the reliability of radar by improving the angular resolution. And a lightweight algorithm with a small memory footprint is desired considering that limited computational resources are accessible for automotive radar. Conventionally, iterative algorithms such as iterative shrinkage thresholding algorithm (ISTA) were used for DOA estimation. However, algorithms like ISTA can require many iterations to converge, and a lot of manual parameter tuning is required to obtain optimal performance. Learned ISTA (LISTA) has been used to approximate ISTA with fewer iterations without the necessity of manual tuning by unfolding the iterative algorithm as a neural network. But directly using LISTA is not suitable for DOA estimation due to the large size of the matrices that need to be learned. The large number of learning parameters require a lot of training data, a long training time, and heavy computation. This work proposes to use circular convolutions to reduce the number of learning parameters in the model as well as computation. We show that the circular convolution-based ISTA has better performance metrics than the traditional ISTA.

This paper proposes a novel framework for analyzing the localization accuracy of data fusion for fingerprinting approaches in non-line-of-sight (NLOS) environments. Using simulation data generated for two very different NLOS environments (a suburban area of $3.3km\times 3.3km$ in Santa Clara, California, and a mountainous area of $11.4km\times 11.4km$ in the Caspian region), we establish novel channel models for measurement differences of three data types (received signal strength indicator (RSS), time of arrival (TOA) and direction of arrival (DOA)) at $K$ neighboring nodes of an arbitrary node. The crucial point is that the modeling errors for each of the three data types are shown to be jointly Gaussian distributed. Based on these measurement difference models, Cramer-Rao Lower Bound (CRLB) is used as a benchmark to evaluate $K$ -nearest neighbor (KNN) and Weighted $K$ -Nearest Neighbor (WKNN). It is shown that the proposed CRLB analyses can be employed to evaluate fingerprinting systems with various designs (such as different data types and fusion options) and different configurations (such as densities of reference nodes and numbers of anchor nodes) in diverse NLOS environments(such as suburban and mountainous regions).

This paper provides an analysis of radio wave scattering for frequencies ranging from the microwave to the Terahertz band (e.g., 1 GHz - 1 THz), by studying the scattering power reradiated from various types of materials with different surface roughnesses. First, fundamentals of scattering and reflection are developed and explained for use in wireless mobile radio, and the effect of scattering on the reflection coefficient for rough surfaces is investigated. Received power is derived using two popular scattering models - the directive scattering (DS) model and the radar cross section (RCS) model through simulations over a wide range of frequencies, materials, and orientations for the two models, and measurements confirm the accuracy of the DS model at 140 GHz. This paper shows that scattering can become a prominent propagation mechanism as frequencies extend to millimeter-wave (mmWave) and beyond, but at other times can be treated like simple reflection. Knowledge of scattering effects is critical for appropriate and realistic channel models, which further support the development of massive multiple input-multiple output (MIMO) techniques, localization, ray tracing tool design, and imaging for future 5G and 6G wireless systems.

In the era of the Internet of Things and Artificial Intelligence, the Wi-Fi fingerprinting-based indoor positioning system (IPS) has been recognized as the most promising IPS for various applications. Fingerprinting-based algorithms critically rely on a fingerprint database built from machine learning methods. However, currently methods are based on single-feature Received Signal Strength (RSS), which is extremely unstable in performance in terms of precision and robustness. The reason for this is that single feature machines cannot capture the complete channel characteristics and are susceptible to interference. The objective of this paper is to exploit the Time of Arrival (TOA) feature and propose a heterogeneous features fusion model to enhance the precision and robustness of indoor positioning. Several challenges are addressed: (1) machine learning models based on heterogeneous features, (2) the optimization of algorithms for high precision and robustness, and (3) computational complexity. This paper provides several heterogeneous features fusion-based localization models. Their effectiveness and efficiency are thoroughly compared with state-of-the-art methods.

Localization schemes based on direction of arrival (DOA) in none-line-of-sight (NLOS) environments are developed. The proposed kernel-based machine learning method is innovative and can provide accurate position estimation under none-line-of sight (NLOS) conditions. The proposed kernel-based method is compared with the Weighted K-nearest neighborhood (WKNN) fingerprinting method using simulated DOA data in practical rural environment. It shows that the kernel-based method gives more accurate localization results.

Currently there is a trend in indoor localization by utilizing machine learning. However, the precision and robustness are limited due to single feature machine learning scheme. The reason behind is that single feature cannot capture the complete channel characteristics and susceptible to interference. The objective of this paper is to introduce heterogeneous features fusion model to enhance the precision and robustness of indoor positioning. Its effectiveness and efficiency are proved by comparing with current benchmark.