Zeitschriftenaufsätze

In this paper, we present an analytical performance assessment of 2-D Tensor ESPRIT in terms of physical parameters. We show that the error in the r -mode depends only on two components, irrespective of the dimensionality of the problem. We obtain analytical expressions in closed form for the mean squared error (MSE) in each dimension as a function of the signal-to-noise (SNR) ratio, the array steering matrices, the number of antennas, the number of snapshots, the selection matrices, and the signal correlation. The derived expressions allow a better understanding of the difference in performance between the tensor and the matrix versions of the ESPRIT algorithm. The simulation results confirm the coincidence between the presented analytical expression and the curves obtained via Monte Carlo trials. We analyze the behavior of each of the two error components in different scenarios.

In this paper, we consider the channel estimation problem in sub-6 GHz uplink wideband multiple-input multiple-output (MIMO)-orthogonal frequency-division multiplexing (OFDM) communication systems, where a user equipment with a fully-digital beamforming structure is communicating with a base station having a hybrid analog-digital (HAD) beamforming structure. A novel channel estimation method called Sequential Alternating Least Squares Approximation (SALSA) is proposed by exploiting a hidden tensor structure in the uplink measurement matrix. Specifically, by showing that any MIMO channel matrix can be approximately decomposed into a summation of R factor matrices having a Kronecker structure, the uplink measurement matrix can be reshaped into a 3-way tensor admitting a Tucker decomposition. Exploiting the tensor structure, the MIMO channel matrix is estimated sequentially using an alternating least squares (ALS) method. Detailed simulation results are provided showing the effectiveness of the proposed SALSA method as compared to the classical least squares and linear minimum mean squared-error (LMMSE) methods.

In the last decade, the signal processing (SP) community has witnessed a paradigm shift from model-based to data-driven methods. Machine learning (ML) - more specifically, deep learning - methodologies are nowadays widely used in all SP fields, e.g., audio, speech, image, video, multimedia, and multimodal/multisensor processing, to name a few. Many data-driven methods also incorporate domain knowledge to improve problem modeling, especially when computational burden, training data scarceness, and memory size are important constraints.

Multidimensional channel sounding measures the geometrical structure of mobile radio propagation. The parameters of a multipath data model in terms of directions, time-of-flight, and Doppler shift are estimated from observations in frequency, time, and space. A maximum likelihood estimation framework allows joint high resolution in all dimensions. The prerequisite for this is an appropriate parametric data model that represents the multipath propagation correctly. At the same time, a device data model is necessary that typically results from calibration measurements. The used model should be as simple as possible, since its structure has a considerable effect on the estimation effort. For instance, the inherent effort in parameter search is reduced if the influence of the parameters is kept independent. Therefore, the data model is characterized by several approximations. The most important is the “narrowband assumption,” which assumes a low relative bandwidth and also avoids considering any frequency response in magnitude and phase. We extend the well-known multidimensional Richter maximization approach (RIMAX) parameter estimation framework by including proper frequency responses. The advantage reveals itself with high bandwidth in the mmWave and sub-THz range. It allows for a more realistic modeling of antenna arrays, and it breaks with the usual narrowband model and allows a better modeling of mutual coupling and time delay effects. If the interacting object extends over several delay bins (hence, an extended target in radar terminology), we propose a model that assigns a short delay spread and a frequency response to the propagation path that associates it with the respective object. We verify the validity of the device model by numerical experiments on simulated and measured antenna data and compare it with RIMAX. In addition, we use synthetic data based on ray-tracing results and measurements both ranging from 27.0 to 33 GHz with known ground-truth information and show that the proposed estimator delivers better performance for higher relative bandwidths than the conventional RIMAX implementation.

The extension of low-earth orbit (LEO) services to non-terrestrial mobile communications has huge potential for eliminating network white spots and providing high-speed, low-latency links with worldwide geographic coverage. State-of-the-art user terminals for mobile platforms are too large for integration into a passenger vehicle. Antenna elements loaded with a dielectric superstrate could potentially lead to a considerable miniaturization of the user terminal. As per link budget calculations, an array with a gain of 27 dBi is necessary to ensure a throughput of 25 Mbps in the downlink at the Ku-band. A conventional array with a gain of 6 dBi per element, assuming a 12 × 12 arrangement with half-wavelength spacing, would require a footprint of 36 λ2 at 10 GHz to achieve this target and appears unsuitable for automotive integration. This paper proposes a low-profile, dual-band, dual-polarized, vertically stacked patch antenna with superstrate loading and shows that the inclusion of the superstrate improves the antenna’s gain by at least 3 dB. Therefore, compared to a conventional array, a superstrate-loaded array would need only half of the number of elements to meet the target gain, thus occupying only half of the surface area, and offers better integration for automotive applications. Requiring half of the number of elements also implies considerably reduced design complexity and cost.