检索结果-内蒙古大学图书馆

Reducing Unnecessary Handovers Using Logistic Regression in 5G Networks

IEEE ACCESS 2025年 13卷 78707-78726页

作者： Fernandes, Alison M. Del Monego, Hermes I. Chang, Bruno S. Munaretto, Anelise Univ Tecnol Fed Parana UTFPR CPGEI BR-80230901 Curitiba Brazil

The evolution of wireless communication has brought great benefits to society, such as multi-connectivity, increased connection speed, low latency, and elevated throughput. However, it has also raised concerns regarding security and the procedures of network swaps, more known as handovers. This paper proposes using machine learning in 5G mobile networks, employing a Logistic Regression algorithm to predict handovers. Considering selection criteria like Reference Signal Received Power, Signal-to-Interference-Plus-Noise Ratio, Received Signal Strength Indicator, and distance, our proposal shows that it can significantly reduce the consequences of frequent handovers and provide a better quality of service to the user, together with low computational complexity. The presented results prove the viability of this proposal.

关键词： Handover 5G mobile communication Computer architecture Microprocessors machine learning machine learning algorithms Long short term memory Throughput Base stations Proposals machine learning logistic regression supervised learning 5G network OMNeT plus plus Simu5G user equipment reference signal received power signal-to-interference-plus-noise ratio received signal strength indicator radio link control throughput 3GPP

来源：评论

学校读者我要写书评

暂无评论

Distributed Optimization for Quadratic Cost Functions With Quantized Communication and Finite-Time Convergence

引用

IEEE TRANSACTIONS ON CONTROL OF NETWORK SYSTEMS 2025年第1期12卷 930-942页

作者： Rikos, Apostolos I. Grammenos, Andreas Kalyvianaki, Evangelia Hadjicostis, Christoforos N. Charalambous, Themistoklis Johansson, Karl H. KTH Royal Inst Technol Div Decis & Control Syst S-11428 Stockholm Sweden Digital Futures S-11428 Stockholm Sweden Univ Cambridge Dept Comp Sci & Technol Cambridge CB3 0FD England Alan Turing Inst London NW1 2DB England Univ Cyprus Sch Engn Dept Elect & Comp Engn CY-1678 Nicosia Cyprus Aalto Univ Sch Elect Engn Dept Elect Engn & Automat Espoo 02150 Finland

In this article, we propose two distributed iterative algorithms that can be used to solve the distributed optimization problem for quadratic local cost functions over large-scale networks in finite time. The first algorithm exhibits synchronous operation while the second one exhibits asynchronous operation. Both algorithms operate exclusively with quantized values. This means that the information stored, processed, and exchanged between neighboring nodes is subject to deterministic uniform quantization. The algorithms rely on event-driven updates in order to reduce energy consumption, communication bandwidth, network congestion, and/or processor usage. Finally, once the algorithms converge, nodes distributively terminate their operation. We prove that our algorithms converge in a finite number of iterations to the exact optimal solution depending on the quantization level, and we present applications of our algorithms to, first, optimal task scheduling for data centers, and second, global model aggregation for distributed federated learning. We provide simulations of these applications to illustrate the operation, performance, and advantages of the proposed algorithms. In addition, it is shown that our proposed algorithms compare favorably to algorithms in the current literature.

关键词： Convergence Quantization (signal) Stochastic processes Task analysis machine learning algorithms Cost function Control systems Distributed algorithms federated learning finite-time optimization quantization resource allocation

来源：评论

学校读者我要写书评

暂无评论

Adaptive Phase Shifters for Hybrid Beamforming in mmWave Systems

引用

IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS 2025年第2期24卷 1104-1116页

作者： Alouzi, Mohamed Yanikomeroglu, Halim Kurt, Gunes Karabulut Carleton Univ Dept Syst & Comp Engn Ottawa ON K1S 5B6 Canada Polytech Montreal Polygrames Res Ctr Dept Elect Engn Montreal PQ H3T 1J4 Canada

Full-array (FA) hybrid beamforming, integrating analog and digital components, is employed in millimeter wave (mmWave) systems to achieve directional signal transmission with enhanced gains. The main hardware challenge in this hybrid configuration resides in the analog segment, where each antenna at the transmitters and receivers is linked to an entire network of phase shifters (PSs). To address this, researchers have investigated sub-array (SA) hybrid designs that use fewer PSs to reduce hardware complexity;however, the number of PSs scales linearly with the number of antenna subarrays. This paper introduces innovative adaptive phase shifters (APSs) designed for hybrid beamforming that feature low hardware complexity and operate independently of the number of antenna arrays and radio frequency (RF) chains. The proposed APSs hybrid scheme is designed to achieve performance comparable to the FA iterative hybrid design while minimizing the number of required PSs. Specifically, the number of PSs can vary from two per RF chain to the entire network of PSs, allowing for a more efficient and high-performance design of hybrid systems. Furthermore, the full network of PSs per RF chain, corresponding to the total number of antenna elements, is reduced using modified K-means algorithms. Simulation results demonstrate that the spectral efficiency performance of APSs hybrid designs using only two PSs surpasses that of conventional FA hybrid design and SA hybrid design and exhibits similar performance to FA iterative hybrid designs with a smaller number of PSs. The proposed APSs hybrid design represents promising technology for 6G systems and beyond, owing to its innovative design with low hardware complexity and high-gain spectral efficiency.

关键词： Radio frequency Array signal processing Antenna arrays Hardware Spectral efficiency Hybrid power systems Millimeter wave communication Antennas Power demand Phased arrays Full-array and sub-array hybrid precoding/combing design mmWave systems massive MIMO systems machine learning algorithms

来源：评论

学校读者我要写书评

暂无评论

A Rapid-Accurate Fault Diagnosis Method Based on Cumulative Probability Distribution for Lithium-Ion Battery Packs

引用

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS 2025年第4期72卷 3896-3904页

作者： Zhang, Zhen Gu, Xin Mao, Ziheng Li, Jinglun Li, Xiangjun Shang, Yunlong Shandong Univ Sch Control Sci & Engn Jinan 250061 Peoples R China China Elect Power Res Inst Beijing 100192 Peoples R China

The iterative innovation and development of fault diagnosis methods have attracted more and more attention as a crucial technology in battery management systems. Nevertheless, the anomalous characteristics associated with early faults are not obvious, which are challenging to identify through conventional diagnosis techniques. For this reason, this article proposes a rapid-accurate fault diagnosis method based on cumulative probability distribution (CPD) for lithium-ion battery packs. The CPD algorithm can transform the battery voltage sequence into a nontime series. Two fault diagnosis submethods are designed based on the CPD algorithm, including rapid predetection method A with long-term voltage data as input, and accurate diagnosis method B with short-term voltage as input. Thereafter, these proposed methods maintain a balance between diagnosis efficiency and accuracy. The experimental results demonstrate that the proposed method can detect battery early faults and estimate the occurrence time. More importantly, the high fault detection rate (99%) and detection accuracy rate (98.02%) validate the effectiveness and universality. The research findings in this article have excellent development prospects and highlight the application potential of probability distribution for battery fault diagnosis.

关键词： Fault diagnosis Accuracy Lithium-ion batteries Fault detection machine learning algorithms Threshold voltage Windows Voltage control Time series analysis Probability distribution Accurate diagnosis cumulative probability distribution fault diagnosis lithium-ion battery rapid predetection

来源：评论

学校读者我要写书评

暂无评论

Laplace Distribution Based Robust Identification of Errors-in-Variables Systems With Outliers

引用

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS 2025年

作者： Guo, Fan Liu, Kun Huang, Biao Nanjing Inst Technol Sch Automat Nanjing 211167 Peoples R China Univ Alberta Dept Chem & Mat Engn Edmonton AB T6G 2G6 Canada

This article considers a robust identification problem of errors-in-variables (EIVs) system with nonlinear input generation process. In addition, we consider that the output data has both measurement noise and disturbance with outliers. The disturbance is described by a Laplace distribution to account for the outliers. Simultaneously, the input data is also contaminated by measurement noise. The parameters and posterior distributions of the proposed model are estimated by utilizing the expectation maximization algorithm as well as particle filter. Since the integrals of posterior state estimation cannot be derived analytically due to nonlinearity, a bank of weighted particles is utilized to approximate the states. The efficacy of the proposed model is demonstrated by simulation and experimental studies. Finally, the proposed robust modeling method is compared with three other robust modeling methods through comparison of mean absolute error (MAE) and mean relative error (MRE) values. The results demonstrate that the proposed method achieves smaller MAE and MRE, indicating its superior robustness in handling both noisy input-output data and outlier-corrupted outputs.

关键词： Noise Mathematical models Noise measurement Data models Gaussian distribution machine learning algorithms Bayes methods Inference algorithms Pollution measurement Approximation algorithms Expectation maximization algorithm Laplace distribution nonlinear errors-in-variables (EIVs) system particle approximation robust identification

来源：评论

学校读者我要写书评

暂无评论

Application of Phi (Φ), the Golden Ratio, in Computing: A Systematic Review

引用

IEEE ACCESS 2025年 13卷 4621-4651页

作者： Akhtaruzzaman, M. Tanvin, Jamal Uddin Shafie, Amir Akramin Shahryer, Fahim Halder, Sachitra MIST Dept Comp Sci & Engn Dhaka 1216 Bangladesh QRARG Dhaka 1216 Bangladesh IIUM Dept Mechatron Engn Kuala Lumpur 53100 Malaysia

The Golden Ratio (Phi ) has intrigued researchers for centuries due to its widespread presence in both nature and science. This article systematically examines Phi 's diverse applications in computing, focusing on its role in enhancing visual appeal, optimizing algorithms, system design, and influencing fields such as biometrics, machine intelligence, and engineering. Through a systematic review of recent studies, this work aims to identify how Phi can improve computational efficiency across various computing domains. The methodology involves a comprehensive analysis of cutting-edge research sourced from 12 reputable digital platforms to explore Phi 's potential. This study highlights the versatility of Golden Ratio, as its applications and relationships were found to account for 51% and 49%, respectively, across a total of 51 selected studies from 27 countries spanning six continents. The review also categorizes findings by their specific applications and contexts, providing valuable insights into Phi 's impact on mathematics, cryptography, search algorithms, machine learning, artificial intelligence, photonics, natural sciences, system design, power engineering, robotics, and practical human life. This study opens promising avenues for future research on the applications of Phi in diverse fields of science, engineering, and computing.

关键词： Databases Systematic literature review machine learning Art Visualization Photonics machine learning algorithms Libraries Cryptography Computer science Golden ratio phi phi in computing Fibonacci number Golden Section Phi applications

来源：评论

学校读者我要写书评

暂无评论

Network-Wide QoT Estimation With Optimized Gradient Transfer Between Wavelengths

引用

JOURNAL OF LIGHTWAVE TECHNOLOGY 2025年第10期43卷 4582-4590页

作者： Mayer, Kayol S. Soares, Jonathan A. Arantes, Dalton S. Rothenberg, Christian E. Mello, Darli A. A. Univ Estadual Campinas Unicamp Sch Elect & Comp Engn FEEC BR-13083852 Campinas SP Brazil

In optical networks, assessing the quality of transmission (QoT) of future demands is critical for optimized network operation. Given the recent advances in network telemetry and Big Data processing, there has been a growing trend to leverage the telemetry of existing connections to enhance the QoT estimation for future demands. Currently, QoT estimation techniques are categorized into three classes: physics-informed analytical models, machine learning (ML) models, and hybrid models, combining the first two classes. Hybrid models have the potential to achieve accurate QoT estimation with reasonable explainability. In this work, we propose a hybrid QoT estimation technique based on gradient transfer between wavelengths using a simplified network digital twin. The proposed approach relies on the GN model to create a starting point NDT (greenfield), which is then updated with signal-to-noise ratio (SNR) telemetry from deployed transponders (brownfield) using gradient descent techniques. By assuming that neighboring channels have similar performance, the method enables learning from adjacent lightpaths. We use well-established ML optimizers to enhance the convergence rate and reduce steady-state estimation error. Compared with classical analytical models, simulation results indicate improved accuracy with reasonable explainability in network scenarios with imperfect amplifier parameters.

关键词： Signal to noise ratio Estimation Quality of transmission Training Optical fiber networks machine learning algorithms Digital twins Analytical models Telemetry Optical add-drop multiplexers Quality of transmission (QoT) estimation optical networks gradient descent optimization techniques machine learning (ML)

来源：评论

学校读者我要写书评

暂无评论

Generalization Performance of Empirical Risk Minimization on Over-Parameterized Deep ReLU Nets

引用

IEEE TRANSACTIONS ON INFORMATION THEORY 2025年第3期71卷 1978-1993页

作者： Lin, Shao-Bo Wang, Yao Zhou, Ding-Xuan Xi An Jiao Tong Univ Ctr Intelligent Decis Making & Machine Learning Sch Management Xian 710049 Peoples R China Univ Sydney Sch Math & Stat Sydney NSW 2006 Australia

In this paper, we study the generalization performance of global minima of empirical risk minimization (ERM) on over-parameterized deep ReLU nets. Using a novel deepening scheme for deep ReLU nets, we rigorously prove that there exist perfect global minima achieving optimal generalization error rates for numerous types of data under mild conditions. Since over-parameterization of deep ReLU nets is crucial to guarantee that the global minima of ERM can be realized by the widely used stochastic gradient descent (SGD) algorithm, our results present a potential way to fill the gap between optimization and generalization of deep learning.

关键词： Deep learning Optimization Convergence Approximation algorithms Error analysis machine learning algorithms Analytical models Sparse matrices Kernel Approximation error empirical risk minimization global minima over-parameterization

来源：评论

学校读者我要写书评

暂无评论

Active Multikernel Sparse Representation for Synthetic Aperture Radar Automatic Target Recognition

引用

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT 2025年 74卷

作者： Tong, Xiaobao Wang, Yong Cent South Univ Sch Automat Changsha 410083 Peoples R China

Multikernel sparse representation (SR) algorithms have made much progress in synthetic aperture radar automatic target recognition (SAR ATR) tasks. These algorithms usually combine all the kernels to capture the nonlinear structures of SAR images for recognition, without considering which kernels are effective or not, thereby resulting in limited recognition performance. To tackle this issue, this article develops an active multikernel SR (AMKSR) algorithm, which includes a kernel alignment module (KAM), a weight update module (WUM), and a three-step optimization module (TOM). Specifically, KAM calculates the alignment score of each kernel by exploring inter-view and intra-view differences. Afterward, WUM actively selects a kernel with both a high alignment score and a high recognition score, in which the recognition score is calculated by utilizing the sparse codes and the dictionary representation matrix. Subsequently, WUM updates the weights of all the kernels by using the selected kernel. However, due to the sparse codes and the dictionary representation matrix may not being optimal, the weights of all the kernels updated by WUM may also not be optimal. To this end, TOM alternately optimizes the sparse codes, the dictionary representation matrix, and the weights of all the kernels, with the aim of acquiring the optimal weights of all the kernels. Finally, AMKSR selects the non-zero weights from the optimal weights of all the kernels to identify effective kernels and combine them for recognition. The results indicate that AMKSR outperforms other advanced algorithms across various scenarios.

关键词： Kernel Radar polarimetry Feature extraction machine learning algorithms Synthetic aperture radar Target recognition Indexes Image recognition Sparse matrices Electromagnetic scattering Active multikernel learning alternative optimization automatic target recognition (ATR) sparse representation (SR)

来源：评论

学校读者我要写书评

暂无评论

Optimizing Crop Recommendations With Improved Deep Belief Networks: A Multimodal Approach

引用

IEEE ACCESS 2025年 13卷 31762-31773页

作者： Madhuri, J. Ramana Reddy, B. Kalnoor, Gauri Indiramma, M. Nagarathna, N. Bangalore Inst Technol Dept CSE Bengaluru 560004 India Chaitanya Bharathi Inst Technol Dept CSE Hyderabad 500075 India Manipal Inst Technol Bengaluru Manipal Acad Higher Educ Dept CSE Manipal 576104 India BMS Coll Engn Dept Artificial Intelligence & Data Sci Bengaluru 560019 India BMS Coll Engn Dept Comp Sci & Engn IoT & CS Bengaluru 560019 India

With the increase in world population, agricultural planning is significant to ensure food security. Timely recommendations for crops could be valuable for planning food production and maintaining food sustainability. This proposed work suggests a crop recommendation model considering physical soil characteristics, chemical soil characteristics, climate, and crop characteristics, using Improved Deep Belief Networks (IDBN). For this study, four important Indian crops-rice, maize, finger millet and sugarcane -were taken into account. The soil, weather, and crop datasets were gathered from different sources and subjected to feature selections method to extract optimal features. The crop recommendation model was developed using the proposed IDBN with Gaussian Restricted Boltzmann machines and Ranger Optimizer with selected optimal features as input. Gaussian Restricted Boltzmann machines enable the use of continuous values, and Ranger Optimizer ensures rapid convergence with selected optimal features as input. The findings from the investigation confirm that the suggested IDBN outperforms conventional Deep Belief Networks (DBN). The effectiveness of the crop recommendation model was evaluated using IDBN, and its performance was compared with several other machine learning algorithms.

关键词： Crops Soil Meteorology Predictive models Neural networks machine learning Data models Artificial neural networks Mathematical models machine learning algorithms Deep belief networks restricted Boltzmann machines crop recommendation DBN

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：