Radio-frequency (RF) signal processing has seen significant advancements with the advent of deep learning, providing more accurate and efficient solutions for tasks such as signal classification and generation. However, most existing methods are heavily dependent on large labeled datasets, which are often scarce and costly to obtain in real-world RF environments. Furthermore, these approaches tend to be task-specific, limiting their ability to generalize across various RF applications. To address these challenges, this paper proposes RF-MAE, a self-supervised adaptive frequency masked autoencoder. RF-MAE leverages self-supervised learning (SSL) to capture intrinsic patterns from large-scale unlabeled RF data. Central to RF-MAE is a novel Adaptive Frequency Masked (AFM) strategy, which dynamically masks frequency components based on their energy distribution. Supported by a robust theoretical foundation, AFM ensures the model focuses on the most informative signal components, thereby enhancing generalization across RF tasks. By pretraining on unlabeled data and fine-tuning on specific tasks, RF-MAE significantly reduces the reliance on labeled datasets while improving adaptability across diverse RF signal processing tasks. Experimental results demonstrate that RF-MAE consistently outperforms traditional models, underscoring its potential to generalize across tasks and deliver superior performance in a wide range of RF signal applications.

Radio Frequency Fingerprinting Identification (RFFI) has become a critical technology in the physical-layer security (PLS) field, with deep learning emerging as the dominant approach over the past decade. However, most deep learning-based models rely on the assumption that training and testing data follow an independent and identical distribution (i.i.d.), which often does not hold in real-world scenarios. This mismatch significantly degrades model performance in cross-domain settings, making cross-domain RFFI a challenging task. Traditional unsupervised domain adaptation (UDA) methods attempt to address this issue by jointly optimizing task loss and domain loss which is able to reduce the distribution gap between training and testing data. However, we observe that during training, the gradients of these two losses often conflict, hindering effective optimization and limiting cross-domain performance improvements. To address these challenges, we propose a novel framework, Gradient Collaborative Optimization with Dynamic Weighted Feature Alignment (GCODWFA). Specifically, GCODWFA introduces a novel Gradient Collaborative Optimization (GCO) loss, which explicitly adjusts the gradient interaction between task and domain losses by optimizing their angular relationship. Additionally, it incorporates a Dynamic Weighted Feature Alignment (DWFA) strategy, which dynamically adjusts the layer-specific weights for feature alignment based on the angular similarity of task and domain gradients. Extensive experiments conducted on multiple datasets demonstrate the superiority of GCODWFA over existing methods.

Channel state information (CSI)-based human activity recognition (HAR) receives increasing research interests due to its broad applications such as human-computer interaction, health care, and security surveillance. Deep learning (DL) methods have been widely adopted on CSI-based HAR tasks to extract features automatically, overcoming the complexity and unstableness of manual feature extraction process. However, many DL approaches fail to customize the designed model structure with the input CSI tensor shape, applying DL models recklessly. In addition, some researchers utilize attention mechanism yet independently along temporal, spatial, or frequency dimension. To address these issues, we propose an end-to-end comprehensive residual convolution framework, namely CHA-Sens, for general CSI-based human activity sensing. CHA-Sens consists of several comprehensive residual convolution modules (CRCM) that feature adaptive kernel size and stride, regional parameter-free attention mechanism and shortcut of identity mapping. Extensive experiments are conducted on three public CSI datasets for recognizing both single human activity (SHA) and human-to-human interaction (HHI) to show the superiority of the proposed design over other state-of-the-art benchmarks.

Automatic modulation recognition (AMR) plays a vital role in non-cooperative communication systems, which is an important technological component of blind signal processing. The application of deep learning (DL) methods in the field of modulation recognition has shown tremendous potential, greatly surpassing the performance of traditional methods. Existing DL-based AMR schemes employ modules of convolution and attentions to capture the local patterns and long-range dependencies for improving the performance of modulation classification, e.g., transformer-based methods. However, the complicated structure of existing artificial neural networks incur exceedingly long running time that hinders the practice implementation of these methods. Against this background, for the first time, this paper shows that the commonly adopted convolution and attentions are not necessary modules for the signal recognition. Specifically, we propose a lightweight architecture exploiting the multilayer perceptrons (MLP) for signal modulation classification, named SigMixer, which includes layers of token-mixing MLP and channel-mixing MLP that are interleaved for enabling spatial and temporal interaction. The proposed SigMixer not only has a simpler structure and lower computational complexity than the existing state-of-the-art approaches, but can also achieve a competitive performance with the formers. Experimental results show that compared with state-of-the-art approaches, SigMixer achieves better recognition accuracy, with 1.44%-10.65% improvement. Besides, the recognition accuracy can reach 89.36% for a low signal-to-noise ratio (SNR) = - 2 dB.

Abstract Robust beamforming is a commonly‐used solution to various array processing problems. Recently, with the inevitable and rapid increasing of array size, the complexity issue arising from high‐dimension matrix computations has become an urgent concern in algorithm design. In this paper, an efficient distributed robust beamforming algorithm is developed for large‐scale arrays. Specifically, the problem of robust minimum variance distortionless response (MVDR) beamforming with sidelobe level control is considered. By fitting the worst‐case reformulation of robust MVDR beamforming into the framework of alternating direction method of multipliers, the problem can be iteratively solved in a distributed manner, with each iteration being computed in closed form. Numerical results demonstrate that this design is more efficient than many related approaches.

This paper considers transmit beamforming and space-time adaptive processing (STAP) filter design in airborne radar for clutter suppression. We suppose that there are some strong clutters, which have a different distribution than the clutter in the cell under test (CUT) based on the database containing a geographical information system (GIS). Hence, we devise a novel algorithm for the joint design of transmit beamforming and STAP filter which aims to maximize the signal-to-clutter-plus-noise ratio (SCNR). To tackle the nonconvex problem, we convert the joint optimization problem into an equivalent beamforming optimization problem and find the optimal so-lution based on the minorization-maximization (MM) technique iteratively. Numerical simulations are provided to val-idate the proposed method and demonstrate its high performance.

This paper considers the best channel selection for reduced-dimension space-time adaptive processing (STAP) in the presence of multiple targets. An algorithm based on semidefinite programming (SDP) is proposed to achieve the best channel selection by optimizing the worst case of the detection performance of the different targets. Compared with some existing algorithms which only consider the single-target case, the proposed algorithm can provide considerable performance improvements for detecting several different targets. Simulations are conducted for validating the proposed method and demonstrating their high performance.