Terahertz ultra-massive MIMO (THz UM-MIMO) is envisioned as one of the key enablers of 6G wireless networks, for which channel estimation is highly challenging. Traditional analytical estimation methods are no longer effective, as the enlarged array aperture and the small wavelength result in a mixture of far-field and near-field paths, constituting a hybrid-field channel. Deep learning (DL)-based methods, despite the competitive performance, generally lack theoretical guarantees and scale poorly with the size of the array. In this paper, we propose a general DL framework for THz UM-MIMO channel estimation, which leverages existing iterative channel estimators and is with provable guarantees. Each iteration is implemented by a fixed point network (FPN), consisting of a closed-form linear estimator and a DL-based non-linear estimator. The proposed method perfectly matches the THz UM-MIMO channel estimation due to several unique advantages. First, the complexity is low and adaptive. It enjoys provable linear convergence with a low per-iteration cost and monotonically increasing accuracy, which enables an adaptive accuracy-complexity tradeoff. Second, it is robust to practical distribution shifts and can directly generalize to a variety of heavily out-of-distribution scenarios with almost no performance loss, which is suitable for the complicated THz channel conditions. For practical usage, the proposed framework is further extended to wideband THz UM-MIMO systems with beam squint effect. Theoretical analysis and extensive simulation results are provided to illustrate the advantages over the state-of-the-art methods in estimation accuracy, convergence rate, complexity, and robustness.

The terahertz (THz) frequency band has recently attracted considerable attention in wireless communications as potential candidate for providing the necessary high bandwidth for demanding applications. With increasing frequency, however, the communication link becomes more vulnerable to blockage, and pathloss increases. While both effects can be mitigated with the judicious utilization of directional beams and Reconfigurable Intelligent Surfaces (RISs), high directivity could potentially increase the probability of undesired misalignment between the beam that is steered by the RIS, and the user. It is therefore crucial to characterize and understand the stochastic behavior of misalignment in RIS-aided THz links. In this work, beam misalignment in RIS-aided links is studied theoretically and analytical models are derived, the validity of which is verified through numerical calculations. It is demonstrated that there is a distinction in the stochastic behavior of misalignment between pointing errors that occur on the steering plane or normally to the steering plane, with direct consequences on the link robustness on misalignment. The analytical models capture the impact of misalignment under these qualitatively different conditions and provide the necessary tools for assessing the stochastic RIS performance with respect to crucial link parameters, such as the transmitter's beam width, the transmitter-RIS distance, the RIS-receiver distance, and the steering angle of the RIS.

We propose a communication framework suitable for data rate maximization in the (THz) bands using adaptive (OFDM) numerology and carrier aggregation. OFDM is a widely adopted waveform due to the simplicity of its implementation and its effectiveness in combating frequency selectivity when the numerology is carefully chosen. However, it suffers from a multitude of limitations, including phase noise due to local oscillator inaccuracies, high peak-to-average power ratio, and is particularly sensitive to time-frequency synchronization errors, which can considerably impact its performance. This is especially relevant at THZ frequencies where larger-than-usual bandwidth is available, and the choice of the numerology should be carefully made given the intrinsic transceiver constraints. Moreover, the abundance of frequency resources in the THZ band imposes new design challenges that should be addressed, especially since the bandwidth usability at these frequencies depends on the communication distance. Hence, we propose a dynamic OFDM numerology adaptation mechanism, where the bandwidth of a CC covered by a single OFDM waveform is changed. For each (CC), the Component Carrier Data Rate (CCDR) is evaluated while considering the effect of both hardware impairments and the wireless channel statistics. We further propose the adoption of a dynamic distance-aware CC allocation such that the available frequency resources are fully utilized, and maximize an (ADR) through the aggregation of several CCs. Simulation results show that the proposed approach yields the highest ADR out of all possible setups.

The affordable hardware cost of ultra-large (XL) reconfigurable intelligent surfaces (RIS) renders them attractive solutions for the performance enhancement of localization and communication systems. However, XL-RIS results in near-field propagation channels, especially for the high-frequency terahertz (THz) communication system, which poses significant challenges for localization and channel estimation. In this paper, we focus on the spherical wavefront propagation in the near field of the THz system with the assistance of a RIS. A near-field channel estimation and localization (NF-JCEL) algorithm is proposed based on the derived second-order Fresnel approximation of the near-field channel model. To be specific, we carefully devise a down-sampled Toeplitz covariance matrix, which enables the decoupling and separate estimation of user equipment (UE) distances and angles of arrival (AoAs). Using the sub-space based method and one-dimensional search, we estimate the angles of arrival (AoAs) and user equipment (UE) distances. The channel attenuation coefficients are obtained through the least square (LS) method. To alleviate the impact of THz channel fading peaks caused by molecular absorption, estimates on multiple sub-bands are utilized for location estimation. Simulation results validate the superiority of the proposed NF-JCEL algorithm to the conventional far-field algorithm and show that higher resolution accuracy can be obtained by the proposed algorithm.

A simultaneously transmitting and reflecting surface (STARS) aided terahertz (THz) communication system is proposed. A novel power consumption model is proposed that depends on the type and resolution of the STARS elements. The spectral efficiency (SE) and energy efficiency (EE) are maximized in both narrowband and wideband THz systems by jointly optimizing the hybrid beamforming at the base station (BS) and the passive beamforming at the STARS. 1) For narrowband systems, independent phase-shift STARSs are investigated first. The resulting complex joint optimization problem is decoupled into a series of subproblems using penalty dual decomposition. Low-complexity element- wise algorithms are proposed to optimize the analog beamforming at the BS and the passive beamforming at the STARS. The proposed algorithm is then extended to the case of coupled phase-shift STARS. 2) For wideband systems, the spatial wideband effect at the BS and STARS leads to significant performance degradation due to the beam split issue. To address this, true time delayers (TTDs) are introduced into the conventional hybrid beamforming structure for facilitating wideband beamforming. An iterative algorithm based on the quasi-Newton method is proposed to design the coefficients of the TTDs. Finally, our numerical results confirm the superiority of the STARS over the conventional reconfigurable intelligent surface (RIS). It is also revealed that i) there is only a slight performance loss in terms of SE and EE caused by coupled phase shifts of the STARS in both narrowband and wideband systems, and ii) the conventional hybrid beamforming achieves comparable SE performance and much higher EE performance compared with the full-digital beamforming in narrowband systems but not in wideband systems, where the TTD-based hybrid beamforming is required for mitigating wideband beam split.

Terahertz (THz) band transmission has the potential to revolutionize future-generation wireless networks by jointly meeting communication and non-communication demands of their connected devices. Recent advances in THz semiconductor technologies and antenna design are closing the THz band gap in millimeter-scale devices which are often battery-limited. The localization of these mobile devices in future wireless networks is of paramount significance for beamforming to overcome THz propagation loss. Consequently, prospective signal processing techniques with co-design architectures are emerging either for joint communication and sensing or for simultaneous information and power transfer. In this paper, we present a novel approach toward Simultaneous Terahertz Imaging with Information and Power Transfer (STIIPT) from a base station transceiver to an Integrated Receiver (IntRx). Leveraging the proposed non-linear rectenna model for energy harvesting, a customized On-Off Keying (cOOK) modulation scheme is proposed for simultaneous communication through a non-linear THz channel while generating a radar-like image to localize the IntRx acting as a target. The proposed theoretical models are corroborated with simulations using a THz band GaAs Schottky diode to demonstrate STIIPT performances which also emphasize the significance of ranging information to optimize rate–energy transfer tradeoff under the cOOK modulation scheme.

This paper presents a comprehensive study for analyzing and modeling the scattering phenomenon of terahertz (THz) waves bouncing on rough surfaces. More specifically, a generic parametric methodology to accurately model the rough characteristics focusing on the root-mean-square (RMS) height and correlation length of a surface is proposed. Firstly, a Monte Carlo-based approach is developed for the efficient and accurate software realization of such rough surfaces. Then, the full-wave simulator FEKO is used to simulate at 300 GHz the far-field behavior of scattering waves on 30 distinct surfaces having 6 RMS heights and 5 correlation lengths. By employing the directive scattering (DS) model, the variation of the scattering coefficient, <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$S$</tex-math></inline-formula> , and its equivalent roughness, <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\alpha _{R}$</tex-math></inline-formula> , are studied for the scattering amplitude aspect. Furthermore, the statistical characteristics of the phase and polarization of the scattered signals are obtained and compared with those available in existing 3GPP channel model standards which are valid for lower frequency bands. These comparisons show that although the phases follow the normal distribution, as is the case with equivalent 3GPP channel models, the cross-polarization ratios (XPRs) follow the logistic distribution. This is an important difference which could have a great impact on the standardization activities of the current 3GPP-related working groups. Finally, the proposed scattering model at the THz band is realized through an algorithmic implementation. Compared with the classical two-dimensional DS model which characterizes only the amplitude of the scattered signal, the proposed three-dimensional scattering model can be efficiently used to accurately characterize all the important parameters of the scattered signals, namely, amplitude, phase, and polarization at the THz band. Additionally, the proposed scattering model can effectively work in conjunction with ray-tracing (RT) schemes leading to precise scattering and channel modeling for new applications, such as holographic radios and multi-user multiple-input multiple-output (MU-MIMO) systems.

This paper studies the channel estimation for wideband multiple-input multiple-output (MIMO) systems equipped with hybrid analog/digital transceivers operating in the millimeter-wave (mmWave) or terahertz (THz) bands. By exploiting the low-rank property of the concatenated channel matrix of the delay taps, we formulate the channel estimation problem as a low-rank matrix sensing (LRMS) problem and solve it using a low-complexity generalized conditional gradient-alternating minimization (GCG-ALTMIN) algorithm. This LRMS-based solution can accommodate different precoder/combiner and training structures. In addition, it does not require knowledge about the array responses at the transceivers, in contrast to most existing solutions allowing low training overhead. Furthermore, a preconditioned conjugate gradient (PCG) algorithm-based implementation and a low-rank matrix completion (LRMC) formulation are proposed to further reduce the computational complexity. In order to enhance the channel estimation performance for fat and tall channel matrices, we introduce a matrix reshaping approach that can preserve the channel rank by exploiting the shift-invariance property of uniform arrays. We also introduce a spectrum denoising (SD) approach for further improving the performance when the array responses are known and the number of paths is small. These approaches can effectively enhance the performance at a given training overhead. Simulation results suggest that the proposed solutions can achieve higher channel estimation accuracy and reduce the computational complexity as compared to several representative channel estimation schemes.