Simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS) represents a novel technology that extends communication coverage to the entire space, compared to the conventional reconfigurable intelligent surface (RIS) covering only half space. However, previous studies mostly focused solely on best-effort services, potentially resulting in lower communication quality for users with poor channel conditions or even communication failure. Consequently, the fairness among users is investigated in the STAR-RIS-assisted multi-input-multi-output system to improve the system performance. To minimize the maximum average mean-square error of each user, a fair beamforming design is utilized to jointly optimize the transmit precoding matrices, reflecting and transmitting coefficients (RTCs), as well as receive combining matrices. To tackle this problem, we first decompose it into three tractable subproblems and then utilize an alternating-optimization algorithm to optimize each one alternately. For the transmit precoding and receive combining matrices, each matrices are obtained by transforming the nonconvex subproblem into a second-order cone programming problem. For the RTCs, considering a low-cost coupled phase shift STAR-RIS model, an element-wise two-stage search algorithm is proposed to optimize the phase shift and amplitude for both energy splitting and mode switching STAR-RIS protocols. Also, the proposed algorithms are extended to the robust beamforming design under imperfect channel state information. Simulation results demonstrate that the proposed algorithm outperforms the other benchmark schemes in the STAR-RIS-assisted MIMO system. Moreover, its performance is closest to the lower bound under various differences in channel conditions among users.

Optical Coherence Tomography (OCT) is a crucial imaging modality for diagnosing and monitoring retinal diseases. However, the accurate segmentation of fluid regions and lesions remains challenging due to noise, low contrast, and blurred edges in OCT images. Although feature modeling with wide or global receptive fields offers a feasible solution, it typically leads to significant computational overhead. To address these challenges, we propose LKMU-Lite, a lightweight U-shaped segmentation method tailored for retinal fluid segmentation. LKMU-Lite integrates a Decoupled Large Kernel Attention (DLKA) module that captures both local patterns and long-range dependencies, thereby enhancing feature representation. Additionally, it incorporates a Multi-scale Group Perception (MSGP) module that employs Dilated Convolutions with varying receptive field scales to effectively predict lesions of different shapes and sizes. Furthermore, a novel Aggregating-Shift decoder is proposed, reducing model complexity while preserving feature integrity. With only 1.02 million parameters and a computational complexity of 3.82 G FLOPs, LKMU-Lite achieves state-of-the-art performance across multiple metrics on the ICF and RETOUCH datasets, demonstrating both its efficiency and generalizability compared to existing methods.

With the rapid advancement of deep learning, the neural network has become the primary approach for non-profiled side-channel attacks. Nevertheless, challenges arise in practical applications due to noise in collected power traces and the substantial amount of data required for training deep learning neural networks. Additionally, acquiring measuring equipment with exceptionally high sampling rates is difficult for average researchers, further obstructing the analysis process. To address these challenges, in this paper, we propose a novel architecture for non-profiled differential deep learning analysis, employing a hybrid-supervised neural network. The architecture incorporates a self-supervised autoencoder to enhance the features of power traces before they are utilized as training data for the supervised neural network. Experimental results demonstrate that the proposed architecture not only outperforms traditional differential deep learning networks by providing a more obvious distinction, but it also achieves key discrimination with reduced computational costs. Furthermore, the architecture is evaluated using small-scale and downsampled datasets, confirming its ability recover correct keys under such conditions. Moreover, the altered architecture designed for data resynchronization was proved to have the ability to distinguish the correct key from small-scale and desynchronized datasets.

In recent years, the deep learning method has emerged as a mainstream approach to non-profiled side-channel attacks. However, most existing methods of deep learning-based non-profiled side-channel attack rely on traditional metrics such as loss and accuracy, which often suffer from unclear results in practical scenarios. Furthermore, most previous studies have not fully considered the properties of power traces as long time-series data. In this paper, a novel non-profiled side-channel attack architecture is proposed, which incorporates the attention mechanism and derives a corresponding attention metric. By attaching the attention mechanism after the network layers, the attention mechanism provides a quantitative prediction of correct key. Moreover, this architecture can effectively extract and analyze the features from long power traces. The success rate on different datasets is at least 86%, which demonstrates the superior reliability of this architecture compared to other works when facing various countermeasures and noise. Notably, even in scenarios where traditional loss and accuracy metrics fail to provide reliable results, the proposed attention metric remains capable of accurately distinguishing the correct key.

Wideband direction-of-arrival (DOA) estimation is a fundamental problem in array signal processing, with broad applications in acoustic localization and related areas. This paper introduces a weighted multi-frequency framework formulated as a semi-definite program to achieve gridless estimation. The proposed approach couples the non-convex sparsity-promoting objectives of multiple subbands via convex constraints, thereby enhancing group sparsity and enabling effective information fusion across frequencies. The formulation demonstrates robustness against aliasing and the associated collision effect, while maintaining favorable convergence properties. In addition, detailed algorithmic implementations are developed for practical noisy environments, and the feasibility of achieving underdetermined DOA estimation is further investigated. Simulation results validate the theoretical analysis and confirm the superior performance of the proposed method.

Physically unclonable functions (PUFs) are crucial for enhancing cybersecurity by providing unique, intrinsic identifiers for electronic devices, thus ensuring their authenticity and preventing unauthorized cloning. The SRAM-PUF, characterized by its simple structure and ease of implementation in various scenarios, has gained widespread usage. The soft-decision Reed–Muller (RM) code, an error correction code, is commonly employed in these designs. This paper introduces the design of an RM code soft-decision attack algorithm to reveal its potential security risks. To address this problem, we propose a soft-decision SRAM-PUF structure based on the elliptic curve digital signature algorithm (ECDSA). To improve the processing speed of the proposed secure SRAM-PUF, we propose a custom ECDSA scheme. Further, we also propose a universal architecture for the critical operations in ECDSA, elliptic curve scalar multiplication (ECSM), and elliptic curve double scalar multiplication (ECDSM) based on the differential addition chain (DAC). For ECSMs, iterations can be performed directly; for ECDSMs, a two-dimensional DAC is constructed through precomputation, followed by iterations. Moreover, due to the high similarity of ECSM and ECDSM data paths, this universal architecture saves hardware resources. Our design is implemented on a field-programmable gate array (FPGA) and an application-specific integrated circuit (ASIC) using a Xilinx Virtex-7 and an TSMC 40 nm process. Compared to existing research, our design exhibits a lower bit error rate ($2.7\times {10}^{-10}$) and better area–time performance (3902 slices, 6.615 $\mathsf{\mu }$s ECDSM latency).