The Rydberg atomic receiver (RARE) has garnered increasing attention in quantum communication due to its capability for high-precision signal sensing and detection. Recent advancements have led to the integration of RARE into multiple-input multiple-output (MIMO) systems. Signal detection in RARE MIMO systems presents a distinct biased phase retrieval (PR) challenge compared to conventional MIMO detection problems associated with radio frequency (RF) chains, rendering many traditional high-performance MIMO detectors inapplicable. This paper investigates the multiuser RARE MIMO problem under M -ary phase-shift keying (MPSK) modulations. The central challenge is to jointly address the biased PR formulation and the discrete MPSK constellation—an area not extensively explored in existing literature. We develop a custom approach that employs a smoothing technique to alleviate the nonsmoothness in the biased PR objective and a penalty transformation to tackle the discrete MPSK structure. The resulting algorithm combines a Majorization-Minimization (MM) framework with a modified Wirtinger flow (WF) method. Numerical simulations demonstrate that our proposed approach achieves superior detection accuracy compared to state-of-the-art detectors while maintaining lower computational complexity.

Language-queried target sound extraction (TSE) aims to extract specific sounds from mixtures based on language queries. Traditional fully-supervised training schemes require extensively annotated parallel audio-text data, which are labor-intensive. We introduce a parallel-data-free training scheme, requiring only unlabelled audio clips for TSE model training by utilizing the contrastive language-audio pre-trained model (CLAP). In a vanilla parallel-data-free training stage, target audio is encoded using the pre-trained CLAP audio encoder to form a condition embedding, while during testing, user language queries are encoded by CLAP text encoder as the condition embedding. This vanilla approach assumes perfect alignment between text and audio embeddings, which is unrealistic. Two major challenges arise from training-testing mismatch: the persistent modality gap between text and audio and the risk of overfitting due to the exposure of rich acoustic details in target audio embedding during training. To address this, we propose a retrieval-augmented strategy. Specifically, we create an embedding cache using audio captions generated by a large language model (LLM). During training, target audio embeddings retrieve text embeddings from this cache to use as condition embeddings, ensuring consistent modalities between training and testing and eliminating information leakage. Extensive experiment results show that our retrieval-augmented approach achieves consistent and notable performance improvements over existing state-of-the-art with better generalizability.

Designing dual-functional waveforms for integrated sensing and communication (ISAC) under low-resolution hardware constraints remains a significant challenge. In this paper, we propose a novel few-bit multiple-input multiple-output (MIMO) dual-functional radar-communication (DFRC) signal design for uplink-downlink ISAC systems, where the transmitter is equipped with low-resolution digital-to-analog converters (DACs). To mitigate the impact of coarse quantization, we adopt a spatial Sigma-Delta (Σ∆) modulation scheme and formulate an optimization problem to maximize the worst-case signal-to-quantization-plus-noise ratio (SQNR) for target sensing, while satisfying symbol error probability (SEP) constraints for all communication users (CUs). A two-stage solution is proposed: a Σ∆ filter is first optimized, followed by a dual accelerated projected gradient (APG) algorithm for symbol-level precoding to generate the DFRC signal. Simulation results demonstrate the effectiveness of the proposed Σ∆ scheme for suppressing the quantization noise, providing promising performance for both sensing and communication tasks.

The future machine-type communication in internet-of-things (IoT) systems involves a massive number of devices sporadically communicating with a base station (BS) equipped with multiple antennas. Detecting active devices and estimating their associated channels are crucial but challenging due to the large number of potential devices and the small fraction of active devices. Existing studies assume high-resolution analog-to-digital converters (ADCs) at the BS, while there is a growing interest in implementing low-resolution ADCs, particularly one-bit ADCs, in massive multiple-input multiple-output (MIMO) systems. This paper focuses on the joint one-bit active device detection and channel estimation problem. We consider the maximum-likelihood approach and propose a novel expectation maximization (EM) algorithm with acceleration. On the theoretical aspect, we provide the convergent computational complexity analysis for the accelerated EM algorithm. The proposed method, evaluated through numerical simulations, outperforms benchmark algorithms in terms of both estimation accuracy and computational complexity.

Massive multiple-input multiple-output (MIMO) technology has revolutionized wireless communication by significantly enhancing spectral efficiency, however its high energy consumption has become a key concern. There is increasing research interest in implementing massive MIMO systems using low-resolution digital-to-analog converters (DACs) to reduce the hardware cost and energy consumption. Meanwhile, the broadcast nature of wireless communications systems poses security risks, exposing user information to potential eavesdroppers (Eve), and this issue has been studied less in the context of low-resolution massive MIMO systems. This paper investigates the potential of low-resolution massive MIMO systems to enhance physical layer security (PLS) without relying on artificial noise (AN). We propose a novel spatial Sigma-Delta modulation technique that strategically leverages quantization noise to obscure confidential communications from Eve, even with limited channel state information. Our design shifts quantization noise away from legitimate users while maintaining its presence near Eve, thus improving PLS. We formulate the resulting non-convex, semi-infinite design problem and apply a proximal majorization-minimization (PMM) algorithm, ensuring convergence to a Karush–Kuhn–Tucker (KKT) point. To enhance computational efficiency, we introduce a proximal distance algorithm (PDA) that addresses the constraints independently, yielding closed-form solutions for projections and proximal operators. Extensive numerical experiments validate our approach, demonstrating effective noise shaping for both users and Eve. Our findings illustrate that quantization noise can be a valuable asset in securing communications in low-resolution massive MIMO systems.

Federated learning (FL) has been recognized as a viable solution for local-privacy-aware collaborative model training in wireless edge networks, but its practical deployment is hindered by the high communication overhead caused by frequent and costly server-device synchronization. Notably, most existing communication-efficient FL algorithms fail to reduce the significant inter-device variance resulting from the prevalent issue of device heterogeneity. This variance severely decelerates algorithm convergence, increasing communication overhead and making it more challenging to achieve a well-performed model. In this paper, we propose a novel communication-efficient FL algorithm, named FedQVR, which relies on a sophisticated variance-reduced scheme to achieve heterogeneity-robustness in the presence of quantized transmission and heterogeneous local updates among active edge devices. Comprehensive theoretical analysis justifies that FedQVR is inherently resilient to device heterogeneity and has a comparable convergence rate even with a small number of quantization bits, yielding significant communication savings. Besides, considering non-ideal wireless channels, we propose FedQVR-E which enhances the convergence of FedQVR by performing joint allocation of bandwidth and quantization bits across devices under constrained transmission delays. Extensive experimental results are also presented to demonstrate the superior performance of the proposed algorithms over their counterparts in terms of both communication efficiency and application performance.

Target-speaker automatic speech recognition (ASR) aims to transcribe the desired speech of a target speaker from multi-talker overlapped utterances. Most of the existing target-speaker ASR (TS-ASR) methods involve either training from scratch or fully finetuning a pre-trained model, leading to significant training costs and becoming inapplicable to large foundation models. This work leverages prompt tuning, a parameter-efficient fine-tuning approach, to extend Whisper, a large-scale single-talker ASR model, to TS-ASR. Variants of prompt tuning approaches along with their configurations are explored and optimized for TS-ASR. Experimental results show that prompt tuning can achieve performance comparable to state-of-the-art full training approaches while only requiring about 1% of task-specific model parameters. Notably, the original Whisper’s features, such as inverse text normalization and timestamp tagging, are retained in target-speaker ASR, keeping the generated transcriptions natural and informative.

Active reconfigurable intelligent surface (RIS) is a new RIS architecture that can reflect and amplify communication signals. It can provide enhanced performance gain compared to the conventional passive RIS systems that can only reflect the signals. On the other hand, the design problem of active RIS-aided systems is more challenging than the passive RIS-aided systems and its efficient algorithms are less studied. In this paper, we consider the sum rate maximization problem in the multiuser massive multiple-input single-output (MISO) downlink with the aid of a large-scale active RIS. Existing approaches for handling this problem usually resort to general optimization solvers and can be computationally prohibitive. We propose an efficient block successive upper bound minimization (BSUM) method, of which each step has a (semi) closed-form update. Thus, the proposed algorithm has an attractive low per-iteration complexity. By simulation, our proposed algorithm consumes much less computation than the existing approaches. In particular, when the MIMO and/or RIS sizes are large, our proposed algorithm can be orders-of-magnitude faster than existing approaches.

Universal sound separation (USS) aims to extract arbitrary types of sounds from real-world recordings. This can be achieved by language-queried target sound extraction (TSE), which typically consists of two components: a query network that converts user queries into conditional embeddings, and a separation network that extracts the target sound accordingly. Existing methods commonly train models from scratch. As a consequence, substantial data and computational resources are required to make the randomly initialized model comprehend sound events and perform separation accordingly. In this paper, we propose to integrate pre-trained models into TSE models to address the above issue. To be specific, we tailor and adapt the powerful contrastive language-audio pre-trained model (CLAP) for USS, denoted as CLAPSep. CLAPSep also accepts flexible user inputs, taking both positive and negative user prompts of uni- and/or multi-modalities for target sound extraction. These key features of CLAPSep can not only enhance the extraction performance but also improve the versatility of its application. We provide extensive experiments on 5 diverse datasets to demonstrate the superior performance and zero- and few-shot generalizability of our proposed CLAPSep with fast training convergence, surpassing previous methods by a significant margin. Full codes and some audio examples are released for reproduction and evaluation.

In this paper, we consider the one-bit precoding problem for the multiuser downlink massive multiple-input multiple-output (MIMO) system with phase shift keying (PSK) modulation. We focus on the celebrated constructive interference (CI)-based problem formulation. We first establish the NP-hardness of the problem (even in the single-user case), which reveals the intrinsic difficulty of globally solving the problem. Then, we propose a novel negative ℓ1penalty model for the considered problem, which penalizes the one-bit constraint into the objective by a negative ℓ1-norm term, and show the equivalence between (global and local) solutions of the original problem and the penalty problem when the penalty parameter is sufficiently large. We further transform the penalty model into an equivalent min-max problem and propose an efficient alternating proximal/projection gradient descent ascent (APGDA) algorithm for solving it, which performs a proximal gradient decent over one block of variables and a projection gradient ascent over the other block of variables alternately. The APGDA algorithm enjoys a low per-iteration complexity and is guaranteed to converge to a stationary point of the min-max problem and a local minimizer of the penalty problem. To further reduce the computational cost, we also propose a low-complexity implementation of the APGDA algorithm, where the values of the variables will be fixed in later iterations once they satisfy the one-bit constraint. Numerical results show that, compared to the state-of-the-art CI-based algorithms, both of the proposed algorithms generally achieve better bit-error-rate (BER) performance with lower computational cost.

Probabilistic linear discriminant analysis (PLDA) is commonly used in speaker verification systems to score the similarity of speaker embeddings. Recent studies improved the performance of PLDA in domain-matched conditions by diagonalizing its covariance. We suspect such a brutal pruning approach could eliminate its capacity in modeling dimension correlation of speaker embeddings, leading to inadequate performance with domain adaptation. This paper explores two alternative covariance regularization approaches, namely, interpolated PLDA and sparse PLDA, to tackle the problem. The interpolated PLDA incorporates the prior knowledge from cosine scoring to interpolate the covariance of PLDA. The sparse PLDA introduces a sparsity penalty to update the covariance. Experimental results demonstrate that both approaches outperform diagonal regularization noticeably with domain adaptation. In addition, in-domain data can be significantly reduced when training sparse PLDA for domain adaptation.

Symbol-level precoding (SLP) has received tremendous interest in MIMO communications in recent years. In particular, much attention has been paid to the formulation and optimization aspects. In this paper we contribute to these aspects by drawing a connection between SLP and a seemingly unrelated topic—namely, parameter estimation from quantized data. Specifically, we illustrate that the maximum detection probability formulation for SLP is basically the same as the maximum likelihood estimation formulation for quantized linear regression (QLR). This dual relationship is not just an interesting fundamental result. Using this relationship, we show how the expectation maximization (EM) method for QLR, a popular way to deal with QLR, can be repurposed to perform optimization for SLP. Our numerical results suggest that the accelerated EM method for SLP, as a new possibility, is highly efficient.

In this paper we examine the application of variational inference (VI) to MIMO detection. Our study is motivated by the recent interest in applying machine learning concepts to signal processing. VI is an approach for providing friendly approximations of certain intractable posterior probabilities in statistics, and it has been popularly used in machine learning. In MIMO detection we also have a similar problem; specifically, we want to evaluate the posterior symbol probabilities for detection, but they are computationally too expensive to evaluate when the problem size and/or the constellation size are large. By approximating the discrete symbol prior by a continuous Gaussian mixture model, we show how the notion of VI can be used to derive an iterative MIMO detector. Interestingly, the detector resembles the MMSE detector in structure. The performance of the proposed detector is demonstrated by simulations.

Unit-modulus constrained optimization is frequently encountered in many engineering problems. In our recent study for massive MIMO precoding, we devised a penalty method for unit-modulus optimization. In this paper, we revisit this penalty method in several other unit-modulus applications in signal processing. Moreover, as a new result, we show that the concept of our penalty method can be generalized to handle a much broader class of problems, such as those with semi-orthogonal matrix constraints. The rationale is to relax the constraint set as its convex hull, and at the same time, we add a penalty function to force the solution to be an extreme point—which lies in the original constraint set. We show conditions under which the penalty formulation is equivalent to the original problem. We test the penalty method on classic and one-bit MIMO detection under M-PSK constraints, and on phase retrieval. Simulation results indicate that the penalty method yields competitive performance on the aforementioned applications.

Simplex identification via split augmented Lagrangian (SISAL) is a popularly used algorithm in blind unmixing of hyperspectral images. Developed by José M. Bioucas-Dias in 2009, the algorithm is fundamentally relevant to tackling simplex-structured matrix factorization and, by extension, nonnegative matrix factorization, which have many applications under their umbrellas. In this article, we revisit SISAL and provide new meanings to this quintessential algorithm. The formulation of SISAL was motivated from a geometric perspective, with no noise. We show that SISAL can be explained as an approximation scheme from a probabilistic simplex component analysis framework, which is statistical and is principally more powerful in accommodating the presence of noise. The algorithm for SISAL was designed based on a successive convex approximation method, with a focus on practical utility. It was not known, by analyses, whether the SISAL algorithm has any kind of guarantee of convergence to a stationary point. By establishing associations between the SISAL algorithm and a line search--based proximal gradient method, we confirm that SISAL can indeed guarantee convergence to a stationary point. Our re-explanation of SISAL also reveals new formulations and algorithms. The performance of these new possibilities is demonstrated by numerical experiments.

Adopting one-bit analog-to-digital convertors (ADCs) for massive multiple-input multiple-output (MIMO) implementations has great potential in reducing the hardware cost and power consumption. However, distortions caused by quantization raise great challenges. In MIMO orthogonal frequency-division modulation (OFDM) detection, coarse quantization renders the orthogonal separation among subcarriers inapplicable, forcing us to deal with a problem that has a very large problem size. In this paper we study the expectation-maximization (EM) approach for one-bit MIMO-OFDM detection. The idea is to iteratively decouple the MIMO-OFDM detection problem among subcarriers. Using the perspective of block coordinate descent, we describe inexact variants of the classical EM method for providing more flexible and computationally efficient designs. Simulation results are provided to illustrate the potential of the divide-and-conquer strategy enabled by EM.

This paper studies nonconvex quadratically constrained quadratic program (QCQP), which is known to be NP-hard in general. In the past decades, various approximate approaches have been developed to tackle the QCQP, including semidefinite relaxation (SDR), successive convex approximation (SCA), the variable splitting approach, to name a few. While these approaches are effective under some circumstances, they have to either lift the variable dimension or require a feasible starting point, thereby not suitable for the large-scale QCQP or lack of a feasible starting point. In light of this, this work aims at developing an efficient approach to the QCQP without the above mentioned drawbacks. The crux of our approach is the proximal distance algorithm (PDA), which merges the idea of the penalty method and majorization minimization (MM) to provide an efficient (closed-form) iterative algorithm. To demonstrate the effectiveness of the PDA, we test it on the multicast beamforming applications in wireless communications. Simulation results show that the PDA outperforms state-of-the-art algorithms in terms of delivering a better solution with much less running time.

Massive MIMO using low-resolution digital-to-analog converters (DACs) at the base station (BS) is an attractive downlink approach for reducing hardware overhead and for reducing power consumption, but managing the large quantization noise effect is a challenge. Spatial Sigma-Delta (ΣΔ) modulation is a recently emerged technique for tackling the aforementioned effect. Assuming a uniform linear array at the BS, it works by shaping the quantization noise as high spatial-frequency, or angle, noise. By restricting the user-serving region to be within a smaller angular region, the quan-tization noise incurred by the users can be effectively reduced. We previously showed that, under the one-bit DAC case, the quantization noise can be satisfactorily contained using a simple first-order ΣΔ modulation scheme. In this work we study the potential of spatial ΣΔ modulation in the two-bit DAC case and under second-order modulation. Our empirical results indicate that second-order spatial ΣΔ modulation provides better quantization noise suppression.

This paper considers symbol-level precoding (SLP) for multiuser multi-input single-output (MISO) downlink transmission with the aid of intelligent reflecting surface (IRS). Specifically, by assuming one-bit transmitted signals at the base station (BS), which arises from the use of low-resolution DACs in the regime of massive transmit antennas, a joint design of one-bit SLP at the BS and the phase shifts at the IRS is proposed with a goal of minimizing the worst-case symbol error probability (SEP) of the users under the PSK modulation. This joint design problem is essentially a mixed integer nonlinear program (MINLP). To tackle it, we alternately optimize the one-bit signal and the phase shifts. For the former, a dual of the relaxed one-bit SLP problem is solved by the mirror descent (MD) method with the maximum block improvement (MBI) heuristics. For the latter, the accelerated projected gradient (APG) method is employed to optimize the phases. Numerical results demonstrate that the proposed joint design can attain better SEP performance than the conventional linear precoding and one-bit SLP.

This paper considers symbol-level precoding (SLP) for the multiuser multiple-input single-output (MISO) downlink scenario. By exploiting symbol constellation information, SLP has the ability to achieve much better performance than traditional linear beamforming schemes. In this work, we propose an SLP design formulation under quadrature amplitude modulation (QAM) constellations. The objective of the design is to minimize the peak transmission energy over symbol time slots, while, at the same time, satisfying pre-specified symbol error probability (SEP) requirements of all the users. This kind of design can reduce the energy spread over symbol time. The resulting problem is a large-scale convex problem, and we develop an efficient alternating direction method of multipliers (ADMM) algorithm for the problem. Simulation results demonstrate that our proposed algorithm significantly outperforms some conventional linear beamforming schemes.

This paper considers multiuser massive MIMO downlink transmission, where the base station (BS) employs a massive number of transmit antennas, each equipped with a low-resolution phase shifter, to simultaneously shape desired symbols at user side, after passing through the channels and receive beamforming. This channel-aided shaping technique, known as symbol-level nonlinear precoding, has recently gained considerable attention owing to its high power efficiency and low implementation cost. However, the design of the transmit signal itself is challenging because the restriction of the transmit signals to a discrete constant envelope (DCE) set leads to a discrete optimization problem. In this paper, we adopt a minimum symbol-error probability design criterion for joint optimization of the transmit DCE signal at the BS and the receive beamformers at the users. An alternating minimization method is built for the problem. The design of the transmit DCE signal leverages on a negative square penalty (NSP) method developed in our recent work. The design of receive beamformers can be decoupled among users and updated by non-convex gradient projection independently. Our simulation results show that the bit-error rate performance markedly improves as the number of receive antennas increases.

Massive multiple-input multiple-output (MIMO) has the potential to substantially improve the spectral efficiency, robustness and coverage of mobile networks. However, such potential is limited by hardware cost and power consumption associated with a large number of RF chains. Recently, one-bit quantization is proposed to address this issue by replacing high-resolution digital-to-analog converters (DACs) with one-bit DACs, thereby simplifying the RF chains. Despite low system cost, advanced signal processing techniques are needed to compensate for quantization distortions caused by low-resolution DACs. In this paper, a symbol-error-rate (SER)-based one-bit precoding scheme is proposed to minimize the detection error probability of all users under one-bit constraints. The problem is recast as a continuous optimization problem with a biconvex objective. By applying the block coordinate descent (BCD) method and the FISTA method, we develop an efficient iterative algorithm to obtain a one-bit precoding solution. Simulation results demonstrate its superiority over state-of-the-art algorithms in terms of bit error rate performance in high-order modulation cases.

This paper focuses on an average robust transmit beamforming optimization problem for the multiuser multiple-input-single-output (MISO) downlink scenario. In this problem, the channels are modeled as Gaussian variables with mean zero and with known covariance at the transmitter. The design criterion is to maximize the sum of the users' average rates with respect to the channels, subject to the total transmission power constraint. The challenge of this problem is that the average rate function generally admits a complex expression. Such an issue can be tackled by stochastic approximation (SA) approaches, but SA may require a large number of samples, or iterations, to provide reasonable performance. In this work, a simple deterministic approximation scheme is proposed. First, we propose a closed-form surrogate of the per-user average rate function. The proposed surrogate function is shown to have an approximation accuracy within 0.8314 bits from the true average rate. Then, we utilize the proposed surrogate function to establish an algebraically simple alternating optimization algorithm for the beamforming problem. Simulation results show that the proposed algorithm is computationally much more efficient than an SA-based state-of-the-art algorithm when they are compared under similar sum rate performance levels.