Biometric authentication has become increasingly popular due to its security and convenience; however, traditional biometrics are becoming less desirable in scenarios such as new mobile devices, Virtual Reality, and Smart Vehicles. For example, while face authentication is widely used, it suffers from significant privacy concerns. The collection of complete facial data makes it less desirable for privacy-sensitive applications. Lip authentication, on the other hand, has emerged as a promising biometrics method. However, existing lip-based authentication methods heavily depend on static lip shape when the mouth is closed, which can be less robust due to lip shape dynamic motion and can barely work when the user is speaking. In this paper, we revisit the nature of lip biometrics and extract shape-independent features from the lips. We study the dynamic characteristics of lip biometrics based on articulator motion. Building on the knowledge, we propose a system for shape-independent continuous authentication via lip articulator dynamics. This system enables robust, shape-independent, and continuous authentication, making it particularly suitable for scenarios with high security and privacy requirements. We conducted comprehensive experiments in different environments and attack scenarios and collected a dataset of 50 subjects. The results indicate that our system achieves an overall accuracy of 99.06% and demonstrates robustness under advanced replay attacks and AI deepfake attacks, making it a viable solution for continuous biometric authentication in various applications.

Deep learning (DL) based methods for orthogonal frequency division multiplexing (OFDM) radio receivers demonstrated higher signal detection performance compared to the traditional receivers. However, the existing DL-based models, usually adapted from computer vision, aren’t well suited for wireless communications. These models require high computational resources and memory, and have significant inference delays, limiting their use in resource-constrained settings. Additionally, reducing network size to ease resource demands often leads to notable performance degradation. This paper introduces collaborative networks (CoNet), a novel neural network (NN) architecture designed for OFDM receivers. CoNet uses multiple small ResNet or CNN subnet-works to simultaneously process signal features from different perspectives like capturing channel correlations and interference patterns. These subnetworks fuse their outputs through interaction operations (e.g., element-wise multiplication), significantly enhancing detection performance. Simulation results show CoNet significantly outperforms traditional architectures like residual networks (ResNets) in bit error rate (BER) and reduces inference delay when both nets have the same size and the same computational complexity.

Towards fast, hardware-efficient, and lowcomplexity receivers, we propose a compression-aware learning approach and examine it on free-space optical (FSO) receivers for turbulence mitigation. The learning approach jointly quantize, prune, and train a convolutional neural network (CNN). In addition, we propose to have the CNN weights of power of two values so we replace the multiplication operations bit-shifting operations in every layer that has significant lower computational cost. The compression idea in the proposed approach is that the loss function is updated and both the quantization levels and the pruning limits are optimized in every epoch of training. The compressed CNN is examined for two levels of compression (1-bit and 2-bits) over different FSO systems. The numerical results show that the compression approach provides negligible decrease in performance in case of 1-bit quantization and the same performance in case of 2-bits quantization, compared to the full-precision CNNs. In general, the proposed IM/DD FSO receivers show better bit-error rate (BER) performance (without the need for channel state information (CSI)) compared to the maximum likelihood (ML) receivers that utilize imperfect CSI when the DL model is compressed whether with 1-bit or 2-bit quantization.

Discrete Fourier Transform Spread-Orthogonal Frequency Division Multiplexing (DFT-s-OFDM) is a promising waveform for both access and backhaul communication in modern wireless systems, due to its low peak-to-average power ratio. However, its performance is impacted by hardware impairments, particularly oscillator phase noise (PN). This paper presents tailored solutions for mitigating PN in DFT-s-OFDM systems, with distinct approaches for the access layer and backhaul communication. For the access layer, we use CoDiPhy, a deep learning (DL)-based receiver that jointly performs channel estimation, equalization, and PN compensation via a conditional denoising diffusion model. As CoDiPhy’s complexity increases with larger FFT and constellation sizes typical in backhaul scenarios, we propose a DL-aided (DLA) PN compensation approach to address the PN issue while reducing computational burden. The DLA PN method outperforms traditional linear interpolation (LI) by exploiting information from all received signals within a pilot section. Simulation results show that CoDiPhy achieves near-optimal performance in the access layer, with coded BERs within 0.2 dB (at a 10−6BER) of the ideal LMMSE solution. In the backhaul scenario, the DLA PN scheme significantly outperforms LI, enabling 1024-QAM with less than 0.5% pilot overhead.

Deep learning (DL) methods have emerged as promising solutions for enhancing receiver performance in wireless orthogonal frequency-division multiplexing (OFDM) systems, offering significant improvements over traditional estimation and detection techniques. However, DL-based receivers often face challenges such as poor generalization to unseen channel conditions and difficulty in effectively tracking rapid channel fluctuations. To address these limitations, this paper proposes a hybrid receiver architecture that integrates the strengths of both traditional and neural receivers. The core innovation is a discriminator neural network trained to dynamically select the optimal receiver whether it is the traditional or DL-based receiver according on the received OFDM block characteristics. This discriminator is trained using labeled pilot signals that encode the comparative performance of both receivers. By including anomalous channel scenarios in training, the proposed hybrid receiver achieves robust performance, effectively overcoming the generalization issues inherent in standalone DL approaches.

TeraHertz (THz) communication is considered as a promising technology to satisfy the ever-increasing demand for high-rate services in the next generation of wireless communication systems. However, the high-speed high-resolution analog-to-digital converters (ADCs) of THz systems are power hungry and complex to implement. Low-resolution ADCs are considered as an efficient solution to reduce the cost and complexity of THz systems. Especially, one-bit ADCs are of particular interest because they require only a comparator and an automatic gain control is not needed anymore. In this paper, we investigate a point-to-point THz backhaul communication system wherein one-bit ADCs with temporal oversampling are deployed at the receiver side. The bit-error-rate (BER) performance of the proposed system is evaluated through an end-to-end link level simulation. Specifically, we propose a convolutional neural network (CNN) based receiver which considers not only the nonlinearity caused by the one-bit ADCs but also the correlation between the received samples due to the oversampling. Our results demonstrate that the proposed CNN-based receiver can considerably improve the BER performance especially at moderate signal-to-noise ratio (SNR) values. The idea of far-field digital dithering is also proposed to maintain the BER performance at high-SNR regime.

The utilization of Orbital Angular Momentum (OAM) mode multiplexing offers a promising solution to alleviate the signal processing overhead in line-of-sight communication systems. However, concerns regarding the divergence of OAM beams have raised doubts about its practicality over long transmission distances. In this paper, we propose a novel beam steering strategy aimed at mitigating this issue by ensuring the successful reception of all OAM modes. To accomplish this objective, we present an unsupervised learning-based beam steering approach designed to individually steer each mode, thereby facilitating their reception at the receiver. We refer to this approach as Steered OAM (S-OAM). To mitigate inter-mode interference (IMI) from the steering process, we incorporate an equalizer into both the receiver and the training process. This ensures that our proposed S-OAM system is IMI-aware, improving its performance across steered OAM modes. Using S-OAM provides the flexibility of choosing the number of transmitted modes to accommodate low-capacity scenarios in low signal-to-noise ratio or long transmission distances, where a large number of modes lead to weak streams. Furthermore, the proposed approach leverages codebook-based phase shifters at the transmitter, eliminating the need for real-time calculation of phase shifter values and thereby maintaining a low level of signal processing burden at the transmit side, akin to traditional OAM transmission. Simulation results validate the efficacy of S-OAM, demonstrating its capability to generate multiple equally robust streams at the receiver with a total spectral efficiency reaching up to $0.5 \mathrm{bits} / \mathrm{sec} / \mathrm{Hz}$ from the theoretical upper bound.

This paper introduces a novel Model-Driven Deep Learning Assisted (MDLA) approach for phase noise (PN) estimation in wideband wireless backhaul scenarios. MDLA estimation combines a model-driven foundation for problem formulation with the computational power of Convolutional Neural Networks (CNN) to tackle the derived problem effectively. Exploiting inherent inter-correlations among PN-affected signals, MDLA method incorporates preceding and subsequent received signals in the PN estimation process. Moreover, a dedicated neural network is crafted to capitalize on input dependencies. This hybrid methodology demonstrates superior performance across diverse wideband scenarios, including constellation sizes, pilot overhead ratios, PN levels, and signal-to-noise ratios. The proposed MDLA scheme exhibits robustness in the face of reduced pilot overhead and achieves nearly optimal results in mitigating PN effects. In a 2 GHz bandwidth system with strong PN, MDLA remains within 0.2 dB and 0.5 dB of the PN -free system in 64-QAM and 256-QAM scenarios, respectively, at a coded bit-error-rate of $10^{-6}$.

TeraHertz (THz) communication, with its ultra-high bandwidth, is considered as a promising technology for the next generation wireless communication systems. To reduce the implementation cost and power consumption of THz systems, low-resolution analog-to-digital converters (ADCs) are proposed as a practical solution. In this paper, the achievable information rate of a THz backhaul communication link with low resolution ADCs is investigated in the form of generalized mutual information (GMI). Specifically, two approaches are proposed for GMI performance evaluation: analytical approach using uniform quantization noise model and end-to-end link simulation using a dense neural network (DNN) soft demapper. Our results demonstrate that the GMI performance as high as 4.9 bits/symbol can be achieved by using a 64-point quadrature amplitude modulation (QAM) with two-bit resolution ADCs and up-sampling ratio of four at the transmitter. We also investigate geometric constellation shaping (GCS) in low resolution ADC systems using a NN-aided trainable constellation and end-to-end learning of the proposed system.

This study presents an End-to-End Learning (ETEL) architecture designed for wireless backhaul links, tackling hardware-induced impairments such as Power Amplifier Non-Linearity (PA NL) and phase noise. Leveraging geometric constellation shaping, a fully connected neural network (NN) at the transmitter, and a residual-network-based convolutional neural network at the receiver, our solution effectively handles challenges posed by higher-order modulation and diverse NLs. Through joint transceiver NN training with a weighted loss function accounting for throughput, PA linearization, and spectrum mask requirements, our architecture outperforms conventional schemes in strong NL scenarios, supporting constellation sizes up to 256. This positions the proposed ETEL solution as a promising candidate for future backhaul links operating at mmW or sub-THz bands envisioned for 6G networks.

Free space optical (FSO) communication has the potential to develop ultra-fast data links that can be used in a vari-ety of sixth-generation (6G) applications, including heterogeneous networks with enormous connectivity and wireless backhauls for cellular systems. This paper proposes several low-complexity deep neural networks designed to detect received symbols in FSO systems, eliminating the need for channel state information (CSI). We consider different intensity modulation (IM) schemes (e.g., on-off shift keying (OOK) and pulse amplitude modulation (PAM)) and different FSO systems (e.g., single-input single-output (SISO) and single-input multiple-output (SIMO)). We design the neural network to be able to exploit the temporal correlation of the channels and the common information received at every single photodetector (the case of SIMO). The convolutional neural network (CNN) is designed in a way that it can learn the channels without using pilots and also can jointly combine the received signals and detect the symbols for a wide range of signal-to-noise ratio (SNR) values and different turbulence levels. Numerical results show that the proposed CNN demonstrates a better performance compared to the maximum likelihood approach that utilizes imperfect channel information for symbol detection.

The significant surge in e-commerce sales has made parcel delivery services exceedingly popular. In this paper, we investigate a dynamic parcel pick-up problem where a vehicle with limited capacity collects customers’ parcels on a closed tour. We consider stochastic demand, stochastic service requests, and stochastic service cancellations simultaneously. Aimed at minimizing the total travel distance of the vehicle, we formulate the problem as a Markov decision process and jointly apply state aggregation and approximate dynamic programming (ADP) to overcome the curse of dimensionality in solution solving. First, we introduce the concept of$aggregation$$granularity$, which refers to the number of states that are combined into a single common vector. Based on the cardinalities of the post-decision state’s component sets, we propose a state aggregation method that reduces the size of the state searching space from exponential to quartic. By utilizing this aggregation, our routing policy is represented implicitly by state value functions that are stored in a lookup table in memory. Second, we develop a novel and well-performing ADP-based routing policy that differs from the general ADP-based policy. We derive a lower bound of the post-decision state value function and use it in policy development. Third, during online execution, we apply a rollout algorithm with the lower bound-based policy as its base policy, resulting in an online rollout policy. Finally, through numerical analysis, we examine the impact of$aggregation$$granularity$and the advantages of the online rollout policy.

The inherent nonlinear behavior exhibited by power amplifiers (PAs) in saturation mode is a major impediment in wireless systems to achieve higher power efficiency, and higher spectral efficiency. Although PA behavioral modeling and digital predistortion (DPD) techniques are widely utilized at transmitter to characterize and linearize such distorted nonlinear output of PAs, the nonlinear distortions with memory effects have become more severe due to simultaneous strong crosstalk between multiple PA branches in multiple-input multiple-output (MIMO) arrays. Furthermore, current multi-input behavioral models suffer from a sharp increase in number of coefficients as the number of transmitter path increases in MIMO. In overcoming these challenges, we propose a decomposed cross-correlation based single-input-single-output (CC-SISO) behavioral modeling architecture. The proposed solution utilizes a low-cost, novel cross-correlation based method to estimate and cancel the simultaneous nonlinear and reverse crosstalk from multiple PA branches. Once the crosstalk is mitigated, the MIMO DPD can be implemented with single-input DPD blocks which significantly reduces the complexity. Furthermore, CC-SISO DPD eliminates the requirement for signal feedback paths before and after the PA, and thus reduces overall hardware implementation complexity. Through simulations, we demonstrate that the proposed CC-SISO architecture can reduce the overall complexity of state-of-art multi-input DPD models for MIMO systems.

In this paper, the performance of orbital angular momentum (OAM) mode multiplexing systems is investigated in the presence of oscillator phase noise (PN). First, the OAM spectrum is calculated for an OAM carrying beam that is generated by a uniform circular array (UCA) connected to an imperfect oscillator. It is shown that the OAM mode purity is insensitive to PN. Subsequently, the system model of the UCA-based OAM multiplexing in the presence of PN is presented. Using this model, the signal-to-interference-plus-noise ratio (SINR) and the sum-rate of the system are derived as a function of PN statistics. The significant reduction in the system throughput due to the PN confirms the necessity of an effective PN mitigation scheme. Hence, the system model after applying a generic pilot-based PN mitigation is also investigated, and the SINR and the sum-rate after PN compensation are analytically derived. Simulation results, generated for various PN models, reveal the existing design trade-offs between the pilot overhead and the sum-rate in these systems.

In this paper, we present an algorithm to construct a misalignment-robust beamforming (BF) codebook for orbital angular momentum mode multiplexing (OAM-MM) systems. The BF codebook is constructed for a given set of antenna misalignment (AM) parameters, i.e., the AM type, rotation angles, and displacement values. We use the sorted QR-decomposition of the augmented misaligned channel matrix to design a BF codebook that is applied to received signals using phase shifters only. Subsequently, by employing the successive interference cancellation (SIC), the transmitted symbols corresponding to each OAM mode are estimated. The combination of codebook-based BF in the analog domain and the SIC in the digital domain offers an effective strategy to solve the AM problem without sacrificing the computational savings associated with the application of OAM-MM. Simulation results confirm that by adopting the proposed approach, the sum-rate of the misaligned OAM-MM system approaches the theoretical upper bound.

Microwave backhaul links are characterized by high signal-to-noise ratios permitting spectrally-efficient transmission. The used signal constellation sizes and achievable data rates are typically limited by transceiver impairments, predominantly by phase noise from non-ideal carrier generation. In this article, we propose a new method to improve the data rate over such microwave links. We make use of the fact that adjacent frequency channels are inactive in many deployment scenarios. We argue that additional data can be transmitted in the skirts of the spectral mask imposed on the transmission signal by regulation. To accomplish this task, we present a shaped wideband single-carrier transmission using non-Nyquist pulse shapes. In particular, we design spectrum-skirt filling (SSF) pulse shaping filters that follow the spectral mask response, and perform detection using an accordingly increased sampling frequency at the receiver. We evaluate the achievable information rates of the SSF-based transmission considering practical dispersive channels and non-ideal transmitter and receiver processing. To compensate for phase noise impairments, we derive carrier phase tracking and estimation techniques, and utilize them in tandem with nonlinear precoding which mitigates the intersymbol interference introduced by the non-Nyquist SSF shaping filter. Quantitative performance evaluations show that the proposed system design achieves higher data rates in a dispersive microwave propagation environment with respect to the conventional transmission with Nyquist pulse shaping.

The oscillator phase noise (PN) is one of the bottlenecks toward increasing the modulation order in the OFDM-based wireless backhaul links. This paper presents an algorithm to overcome this barrier through a novel time-domain iterative phase noise mitigation method. In the first step, by leveraging the high signal-to-noise ratio of the backhaul links, the PN estimation problem in the OFDM system is formulated similar to the PN estimation in the single carrier system, for which several solutions are proposed in the literature. Subsequently, a decision directed approach is used to solve this estimation problem, which relies on the linear interpolation of refined estimates of the PN. The proposed algorithm secures a significant improvement in the phase noise suppression of the OFDM systems in the high signal-to-noise-ratio regime. By requiring small pilot overhead while avoiding computationally complex operations in the algorithm's iterations, the proposed method is tailored specially for the OFDM-based wireless backhaul links. Numerical results verify the effectiveness of proposed PN mitigation scheme.