检索结果-内蒙古大学图书馆

HEART: Achieving Timely Multi-Model Training for Vehicle-Edge-Cloud-Integrated Hierarchical Federated Learning

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Yang, Xiaohong Liwang, Minghui Wang, Xianbin Cheng, Zhipeng Hosseinalipour, Seyyedali Dai, Huaiyu Jiao, Zhenzhen School of Informatics Xiamen University Fujian China Department of Control Science and Engineering The National Key Laboratory of Autonomous Intelligent Unmanned Systems Japan Frontiers Science Center for Intelligent Autonomous Systems Ministry of Education Tongji University Shanghai China Department of Electrical and Computer Engineering Western University ON Canada School of Future Science and Engineering Soochow University Suzhou China Department of Electrical Engineering University at Buffalo-SUNY BuffaloNY United States Department of Electrical and Computer Engineering NC State University RaleighNC United States iF-Labs Beijing Teleinfo Technology Company Ltd. CAICT Beijing China

The rapid growth of AI-enabled Internet of Vehicles (IoV) calls for efficient machine learning (ML) solutions that can handle high vehicular mobility and decentralized data. This has motivated the emergence of Hierarchical Federated Learning over vehicle-edge-cloud architectures (VEC-HFL). Nevertheless, one aspect which is underexplored in the literature on VEC-HFL is that vehicles often need to execute multiple ML tasks simultaneously, where this multi-model training environment introduces crucial challenges. First, improper aggregation rules can lead to model obsolescence and prolonged training times. Second, vehicular mobility may result in inefficient data utilization by preventing the vehicles from returning their models to the network edge. Third, achieving a balanced resource allocation across diverse tasks becomes of paramount importance as it majorly affects the effectiveness of collaborative training. We take one of the first steps towards addressing these challenges via proposing a framework for multi-model training in dynamic VEC-HFL with the goal of minimizing global training latency while ensuring balanced training across various tasks—a problem that turns out to be NP-hard. To facilitate timely model training, we introduce a hybrid synchronous-asynchronous aggregation rule. Building on this, we present a novel method called Hybrid Evolutionary And gReedy allocaTion (HEART). The framework operates in two stages: first, it achieves balanced task scheduling through a hybrid heuristic approach that combines improved Particle Swarm Optimization (PSO) and Genetic Algorithms (GA);second, it employs a low-complexity greedy algorithm to determine the training priority of assigned tasks on vehicles. Experiments on real-world datasets demonstrate the superiority of HEART over existing methods. © 2025, CC BY-NC-ND.

关键词： Federated learning

Covert Communications in Active-IOS Aided Uplink NOMA Systems With Full-Duplex Receiver

学校读者我要写书评

暂无评论

IEEE Transactions on Vehicular Technology 2025年

作者： Kang, Xueyu Qi, Nan Lv, Lu Boulogeorgos, Alexandros-Apostolos A. Tsiftsis, Theodoros A. Liu, Hongwu Shandong Jiaotong University School of Information Science and Electrical Engineering Jinan250357 China Nanjing University of Aeronautics and Astronautics Key Laboratory of Dynamic Cognitive System of Electromagnetic Spectrum Space Nanjing210016 China Xidian University State Key Laboratory of Integrated Services Networks Xi'an710071 China Xidian University School of Telecommunications Engineering Xi'an710071 China University of Western Macedonia Department of Electrical and Computer Engineering Kozani50100 Greece University of Thessaly Department of Informatics and Telecommunications Lamia35100 Greece University of Nottingham Ningbo China Department of Electrical and Electronic Engineering Ningbo315100 China

In this paper, an active intelligent omni-surface (A-IOS) is deployed to aid uplink transmissions in a non-orthogonal multiple access (NOMA) system. In order to shelter the covert signal embedded in the superposition transmissions, a multi-antenna full-duplex (FD) receiver is utilized at the base-station to recover signal in addition to jamming the warden. With the aim of maximizing the covert rate, the FD transmit and receive beamforming, A-IOS refraction and reflection beamforming, NOMA transmit power, and FD jamming power are jointly optimized. To tackle the non-convex covert rate maximization problem subject to the highly coupled system parameters, an alternating optimization algorithm is designed to iteratively solve the decoupled sub-problems of optimizing the system parameters. The optimal solutions for the sub-problems of the NOMA transmit power and FD jamming power optimizations are derived in closed-form. To tackle the rank-one constrained non-convex fractional programming of the A-IOS beamforming and FD beamforming, a penalized Dinkelbach transformation approach is proposed to resort to the optimal solutions via semidefinite programming. Numerical results clarify that the deployment of the A-IOS significantly improves the covert rate compared with the passive-IOS aided uplink NOMA system. It is also found that the proposed scheme provides better covert communication performance with the optimized NOMA transmit power and FD jamming power compared with the benchmark schemes. © 1967-2012 IEEE.

关键词： Jamming

Experimental Analysis of a Self-Coherent M-QAM Receiver by Means of Recurrent Optical Spectrum Slicing and Direct Detection

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Sozos, Kostas Da Ros, Francesco Yankov, Metodi P. Deligiannidis, Stavros Sarantoglou, George Mesaritakis, Charis Bogris, Adonis University of West Attica Dept. of Informatics and Computer Engineering. Aghiou Spiridonos Egaleo Athens12243 Greece University of West Attica Dept. of Biomedical Engineering Aghiou Spiridonos Egaleo Athens12243 Greece Dept. of Electrical and Photonics Engineering Technical University of Denmark Kgs. LyngbyDK-2800 Denmark

High order modulation formats constitute the most prominent way for increasing spectral efficiency in transmission systems. Coherent transceivers that support such higher order formats require heavy digital signal processing (DSP), which increases the power consumption of coherent pluggables, well above the intensity modulation and direct detection (IM/DD) counterparts. Self-coherent or phase retrieval methods have emerged as potential solutions, trying to combine the merits of coherent technology with the simplicity of direct detection. In this work, we experimentally demonstrate the reception of quadrature amplitude modulation (QAM) modulation formats based on direct detection aided by the recurrent optical spectrum slicing (ROSS) photonic accelerator, utilizing minimal DSP and low modulator driving voltages. We experimentally demonstrate 32 Gbaud QAM-4/16 for 25 km, 50 km and 75 km in the C-band aided by a linear digital equalization and the use of programmable photonics as recurrent optical spectrum slicers. We showcase successful detection with driving swings below Vπ/3 in contrast to the full swing required by conventional coherent transceivers. We further improve the system performance utilizing geometric constellation shaping. Finally, we explore the potential power consumption improvement for the next-generation 1.6T pluggables, showcasing over 40% reduction with respect to the most lightweight state of the art coherent solutions reported in literature. © 2025, CC BY.

关键词： Quadrature amplitude modulation

Energy Landscape Shaping for Robust Control of Atoms in Optical Lattices

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Weidner, C.A. O'Neil, S.P. Jonckheere, E.A. Langbein, F.C. Schirmer, S.G. Quantum Engineering Technology Laboratories H. H. Wills Physics Laboratory Department of Electrical and Electronic Engineering University of Bristol BristolBS8 1FD United Kingdom Department of Electrical Engineering and Computer Science United States Military Academy West PointNY10996 United States Department of Electrical and Computer Engineering University of Southern California Los AngelesCA90007 United States School of Computer Science and Informatics Cardiff University CardiffCF24 4AG United Kingdom Faculty of Science & Engineering Physics Swansea University SwanseaSA2 8PP United Kingdom

Robust quantum control is crucial for realizing practical quantum technologies. Energy landscape shaping offers an alternative to conventional dynamic control, providing theoretically enhanced robustness and simplifying implementation for certain applications. This work demonstrates the feasibility of robust energy landscape control in a practical implementation with ultracold atoms. We leverage a digital mirror device (DMD) to shape optical potentials, creating complex energy landscapes. To achieve a desired objective, such as efficient quantum state transfer, we formulate a novel hybrid optimization approach that effectively handles both continuous (laser power) and discrete (DMD pixel activation) control parameters. This approach combines constrained quasi-Newton methods with surrogate models for efficient exploration of the vast parameter space. Furthermore, we introduce a framework for analyzing the robustness of the resulting control schemes against experimental uncertainties. By modeling uncertainties as structured perturbations, we systematically assess controller performance and identify robust solutions. We apply these techniques to maximize spin transfer in a chain of trapped atoms, achieving high-fidelity control while maintaining robustness. Our findings provide insights into the experimental viability of controlled spin transfer in cold atom systems. More broadly, the presented optimization and robustness analysis methods apply to a wide range of quantum control problems, offering a toolkit for designing and evaluating robust controllers in complex experimental settings. Copyright © 2025, The Authors. All rights reserved.

关键词： Robust control

EPEE: Towards Efficient and Effective Foundation Models in Biomedicine

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Zhan, Zaifu Zhou, Shuang Zhou, Huixue Liu, Zirui Zhang, Rui Department of Electrical and Computer Engineering University of Minnesota MinneapolisMN United States Division of Computational Health Sciences Department of Surgery University of Minnesota MinneapolisMN United States Institute for Health Informatics University of Minnesota MinneapolisMN55455 United States Department of Computer Science and Engineering University of Minnesota MinneapolisMN United States

Foundation models, including language models, e.g., GPT, and vision models, e.g., CLIP, have significantly advanced numerous biomedical tasks. Despite these advancements, the high inference latency and the "overthinking" issues in model inference impair the efficiency and effectiveness of foundation models, thus limiting their application in real-time clinical settings. To address these challenges, we proposed EPEE (Entropy- and Patience-based Early Exiting), a novel hybrid strategy designed to improve the inference efficiency of foundation models. The core idea was to leverage the strengths of entropy-based and patience-based early exiting methods to overcome their respective weaknesses. To evaluate EPEE, we conducted experiments on three core biomedical tasks—classification, relation extraction, and event extraction—using four foundation models (BERT, ALBERT, GPT-2, and ViT) across twelve datasets, including clinical notes and medical images. The results showed that EPEE significantly reduced inference time while maintaining or improving accuracy, demonstrating its adaptability to diverse datasets and tasks. EPEE addressed critical barriers to deploying foundation models in healthcare by balancing efficiency and effectiveness. It potentially provided a practical solution for real-time clinical decision-making with foundation models, supporting reliable and efficient workflows. Copyright © 2025, The Authors. All rights reserved.

关键词：

Physical Layer Security in AmBC-NOMA Networks with Random Eavesdroppers

学校读者我要写书评

暂无评论

IEEE Internet of Things Journal 2025年

作者： Pei, Xinyue Wang, Xingwei Huang, Min Chen, Yingyang Li, Xiaofan Tsiftsis, Theodoros A. Northeastern University School of Computer Science and Engineering Shenyang110819 China Northeastern University College of Information Science and Engineering Shenyang110819 China Jinan University College of Information Science and Technology Guangzhou510632 China Jinan University School of Intelligent Systems Science and Engineering Zhuhai519070 China University of Thessaly Department of Informatics Telecommunications Lamia35100 Greece University of Nottingham Ningbo China Department of Electrical and Electronic Engineering Ningbo315100 China

In this work, we investigate the physical layer security (PLS) of ambient backscatter communication non-orthogonal multiple access (AmBC-NOMA) networks where non-colluding eavesdroppers (Eves) are randomly distributed. In the proposed system, a base station (BS) transmits a superimposed signal to a typical NOMA user pair, while a backscatter device (BD) simultaneously transmits its unique signal by reflecting and modulating the BS’s signal. Meanwhile, Eves passively attempt to wiretap the ongoing transmissions. Notably, the number and locations of Eves are unknown, posing a substantial security threat to the system. To address this challenge, the BS injects artificial noise (AN) to mislead the Eves, and a protected zone is employed to create an Eve-exclusion area around the BS. Theoretical expressions for outage probability (OP) and intercept probability (IP) are provided to evaluate the system’s reliability-security trade-off. Asymptotic behavior at high signal-to-noise ratio (SNR) is further explored, including the derivation of diversity orders for the OP. Numerical results validate the analytical findings through extensive simulations, demonstrating that both the AN injection and protected zone can effectively enhance PLS. Furthermore, analysis and insights of different key parameters, including transmit SNR, reflection efficiency at the BD, power allocation coefficient, power fraction allocated to desired signal, Eve-exclusion area radius, Eve distribution density, and backscattered AN cancellation efficiency, on OP and IP are also provided. © 2014 IEEE.

关键词： Signal to noise ratio

Real-Time Autoregressive Forecast of Cardiac Features for Psychophysiological Applications

学校读者我要写书评

暂无评论

IEEE Journal of Biomedical and Health informatics 2025年 PP卷 PP页

作者： Yaldiz, Cem O. Lin, David J. Gazi, Asim H. Cestero, Gabriela Chen, Chuoqi Bracken, Bethany K. Winder, Aaron Lynn, Spencer Sameni, Reza Inan, Omer T. Georgia Institute of Technology School of Electrical and Computer Engineering AtlantaGA30332 United States Georgia Institute of Technology Department of Biomedical Engineering AtlantaGA30332 United States Charles River Analytics Inc. BostonMA02138 United States Emory University Department of Biomedical Informatics AtlantaGA30322 United States

Forecasting the near-exact moments of cardiac phases is crucial for several cardiovascular health applications. For instance, forecasts can enable the timing of specific stimuli (e.g., image or text presentation in psycholinguistic experiments) to coincide with cardiac phases like systole (cardiac ejection) and diastole (cardiac filling). This capability could be leveraged to enhance the amplitude of a subject's response, prompt them in fight-or-flight scenarios or conduct retrospective analysis for physiological predictive models. While autoregressive models have been employed for physiological signal forecasting, no prior study has explored their application to forecasting aortic opening and closing timings. This work addresses this gap by presenting a comprehensive comparative analysis of autoregressive models, including various forms of Kalman filter-based implementations, that use previously detected R-peak, aortic opening, and closing timings from electrocardiogram (ECG) and seismocardiogram (SCG) to forecast subsequent timings. We evaluate the robustness of these models to noise introduced in both SCG signals and the output of feature detectors. Our findings indicate that time-varying and multi-feature algorithms outperform others, with forecast errors below 2 ms for R-peak, below 3 ms for aortic opening timing, and below 10 ms for aortic closing timing. Importantly, we elucidate the distinct advantages of integrating multi-feature models, which improve noise robustness, and time-varying approaches, which adapt to rapid physiological changes. These models can be extended to a wide range of short-term physiological predictive systems, such as acute stress detection, neuromodulation sensor feedback, or muscle fatigue monitoring, broadening their applicability beyond cardiac feature forecasting. © 2013 IEEE.

关键词： Blood vessels

Polychromatic Coloring of Tuples in Hypergraphs

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Biniaz, Ahmad De Carufel, Jean-Lou Maheshwari, Anil Smid, Michiel Smorodinsky, Shakhar Stojaković, Miloš School of Computer Science University of Windsor Canada School of Electrical Engineering and Computer Science University of Ottawa Ottawa Canada School of Computer Science Carleton University Ottawa Canada Department of Computer Science Ben-Gurion University of the Negev Be’er Sheva Israel Department of Mathematics and Informatics University of Novi Sad Novi Sad Serbia

A hypergraph H consists of a set V of vertices and a set E of hyperedges that are subsets of V . A t-tuple of H is a subset of t vertices of V . A t-tuple k-coloring of H is a mapping of its t-tuples into k colors. A coloring is called (t, k, f)-polychromatic if each hyperedge of E that has at least f vertices contains tuples of all the k colors. Let fH(t, k) be the minimum f such that H has a (t, k, f)-polychromatic coloring. For a family of hypergraphs H let fH(t, k) be the maximum fH(t, k) over all hypergraphs H in H. Determining fH(t, k) has been an active research direction in recent years. This is challenging even for t = 1. We present several new results in this direction for t ≥ 2. • Let H be the family of hypergraphs H that is obtained by taking any set P of points in 2, setting V := P and E := {d ∩ P : d is a disk in R2}. We prove that fH(2, k) ≤ 3.7k, that is, the pairs of points (2-tuples) can be k-colored such that any disk containing at least 3.7k points has pairs of all colors. We generalize this result to points and balls in higher dimensions. • For the family H of hypergraphs that are defined by grid vertices and axis-parallel rectangles in the plane, we show that fH(2, k) ≤ √ck ln k for some constant c. We then generalize this to higher dimensions, to other shapes, and to tuples of larger size. • For the family H of shrinkable hypergraphs of VC-dimension at most d we prove that fH(d+1, k) ≤ ck for some constant c = c(d). Towards this bound, we obtain a result of independent interest: Every hypergraph with n vertices and with VC-dimension at most d has a (d+1)-tuple T of depth at least n/c , i.e., any hyperedge that contains T also contains nc other vertices. We also present analogous bounds for coloring pairs of points with respect to pseudo-disks in the plane. • For the relationship between t-tuple coloring and vertex coloring in any hypergraph H we establish the inequality 1/e · tk1/t ≤ fH(t, k) ≤ fH(1, tk1/t ). For the special case of k = 2, ref

关键词： Knowledge management

IRS Channel Estimation in Cell-free MIMO Network: A Coalition Formation Guided Federated Learning Approach

学校读者我要写书评

暂无评论

IRS Channel Estimation in Cell-free MIMO Network: A Coalitio...

IEEE Conference on Wireless Communications and Networking

作者： Haoxuan Liu Nan Qi Xiaojie Li Alexandros A. A. Boulogeorgos Theodoros A. Tsiftsis Ming Xiao Juha Röning Nanjing University of Aeronautics and Astronautics Nanjing China National Mobile Communications Research Laboratory Southeast University Nanjing China Department of Electrical and Computer Engineering University of Western Macedonia Kozani Greece Department of Informatics and Telecommunications University of Thessaly Lamia Greece Department of Electrical and Electronic Engineering University of Nottingham Ningbo China Ningbo China School of Electrical Engineering of KTH Royal Institute of Technology Stockholm Sweden Biomimetics and Intelligent Systems Group University of Oulu Finland

ISBN: (数字)9798350368369

ISBN: (纸本)9798350368376

The downlink channel estimation is currently a critical bottleneck for IRS-assisted cell-free multiple input multiple output communication. Conventionally, most studies have employed deep learning methods to estimate the high-dimensional, complex cascaded channels generated by IRS, necessitating data collection from all users for centralized model training, which results in excessively large overheads, and data privacy problems. To tackle this challenge, a federated learning (FL)-based channel estimation framework incorporates coalition formation to guide the formation of FL user groups. We propose a coalition formation-enabled federated learning framework for channel estimation, utilizing a deep reinforcement learning (DRL) approach to intelligently group users into multiple coalitions, thereby improving channel estimation accuracy. Moreover, considering that nodes with similar distances to the base station and similar received signal power have a strong likelihood that they experience similar channel fading, we designed a transfer learning method that incorporates both received reference signal power and distance similarity metrics. The transfer learning technique is designed to accelerate the convergence of DRL-federated learning process. Simulations reveal that the proposed algorithms significantly reduce communication overhead for local users and improve data privacy while maintaining commendable channel estimation accuracy.

关键词： Training Data privacy Accuracy Federated learning Transfer learning Channel estimation Deep reinforcement learning Data models Surface treatment Convergence