检索结果-内蒙古大学图书馆

Comparison of QoS optimisation techniques in adaptive smart device peer-to-peer overlay networks

INTERNATIONAL JOURNAL OF parallel EMERGENT AND distributed systems 2021年第6期36卷 668-686页

作者： Maiti, Ananda Maxwell, Andrew Kist, Alexander Univ Southern Queensland Sch Mech & Elect Engn Toowoomba Qld Australia

Smart device networks form the backbone of many Internet of Things (IoT) applications. The performance of these systems depended on many factors. This includes their ability to connect to each other and network latency. In this paper, two algorithms are discussed that build a Peer-to-peer (P2P) overlay network. The focus is on real-time end-to-end remote instrumentation for IoT applications. P2P overlay networks are used for various purposes from content distribution through to media sharing. In contrast to traditional P2P systems, IoT systems can be highly sensitive to network latency and require low bandwidth. The main contributions of this paper are two algorithms to create overlay networks that assign peer nodes to super-peer nodes. The network contains a large number of smart devices and the key aim is to minimize system wide network latency. Other constraints regarding the nodes ability include the ability to become a super-peer node such as connectivity and limited processing capacities of the smart devices are also considered in context of a P2P remote instrumentation system. Both algorithms create a 3-hop network where the latency is lowered as much as possible with the given set of nodes. The algorithms have been validated through simulation.

关键词： distributed computing smart devices peer-to-peer network assignment problem remote instrumentation data clustering

来源：评论

学校读者我要写书评

暂无评论

Anomaly Detection in Data Plane systems using Packet Execution Paths 2

Anomaly Detection in Data Plane Systems using Packet Executi...

引用

2nd ACM SIGCOMM workshop on Secure Programmable Network Infrastructure (SPIN)

作者： Sanghi, Archit Kadiyala, Krishna P. Tammana, Praveen Joshi, Saurabh IIT Hyderabad Hyderabad India Texas Christian Univ Ft Worth TX 76129 USA

ISBN: (纸本)9781450386371

Programmable data planes provide exciting opportunities to realize fast, accurate, and data-driven control-loop decisions. Many data plane systems have been proposed for handling network dynamics (e.g., congestion, failures) in near real-time. The core of these systems has packet-processing data-plane algorithms that continuously monitor traffic and respond automatically. Despite their benefits, automatic response to network events lead to increase in potential sources of inputs, and hence, increase in attack surface. This paper takes a step towards securing such systems by (1) identifying possible attacks on recently proposed data-driven data-plane systems;and (2) designing a scalable tool for detecting such attacks at run time. Our approach models plausible expected behavior and uses the model as a reference to check whether the system is under attack. We conduct preliminary experiments to demonstrate the feasibility of our detection methodology.

关键词： Programmable networks In-network computing Network security Intrusion Detection systems distributed Denial-of-Service attacks

来源：评论

学校读者我要写书评

暂无评论

Impact of the Measurement time Resolution on Energy Key Performance Indicators for distributed Energy Resources: An Experimental Analysis 11

Impact of the Measurement Time Resolution on Energy Key Perf...

引用

11th IEEE International workshop on Applied Measurements for Power systems (AMPS)

作者： Pasetti, Marco Rinaldi, Stefano Bellagente, Paolo Depari, Alessandro Ferrari, Paolo Flammini, Alessandra Sisinni, Emiliano Univ Brescia Dept Informat Engn Brescia Italy

ISBN: (纸本)9781728169231

The growing presence of Renewable Energy Sources (RESs) is heavily affecting the control and management of distribution grids. Battery Energy Storage systems (BESSs) are increasingly adopted to mitigate the adverse impacts caused by the uncertainty of RES generators. Most of the existing works in the literature aims at using BESSs to maximize the self-consumption of energy provided by RES generators or focuses on the mitigation of voltage and/or frequency drift effects. More recently, some research works also studied the effects that BESSs have on the active power flows at the Point of Common Coupling (PCC) of prosumers. All these studies are based on the definition of proper energy Key Performance Indicators (KPIs) based on heterogeneous information collected from several measurement devices deployed in the field. This also means that the results of such studies could be significantly affected by the quality of data, particularly concerning the time resolution of the measurements. To better highlight this concern, this paper proposes an analysis of the impact of the time resolution of measurements on the evaluation of the absolute power ramp reduction at the PCC of prosumers in presence of BESS installations. The analysis was performed on real data collected from the eLUX lab of the University of Brescia, Italy, by varying the time resolution from 5 s to 15 min. The results of the analysis demonstrated that the time resolution has a great impact on the evaluation of the considered KPI, by leading to completely different performance assessments.

关键词： time resolution Energy Key Performance Indicators Renewable Energy Sources Smart Grid Distribution Grid Battery Energy Storage systems

来源：评论

学校读者我要写书评

暂无评论

distributed Deep Reinforcement Learning-Based Energy and Emission Management Strategy for Hybrid Electric Vehicles

引用

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY 2021年第10期70卷 9922-9934页

作者： Tang, Xiaolin Chen, Jiaxin Liu, Teng Qin, Yechen Cao, Dongpu Chongqing Univ Coll Mech & Vehicle Engn Chongqing 400044 Peoples R China Beijing Inst Technol Sch Mech Engn Beijing 100081 Peoples R China Univ Waterloo Dept Mech & Mechatron Engn Waterloo ON N2L 3G1 Canada

Advanced algorithms can promote the development of energy management strategies (EMSs) as a key technology in hybrid electric vehicles (HEVs). Reinforcement learning (RL) with distributed structure can significantly improve training efficiency in complex environments, and multi-threaded parallel computing provides a reliable algorithm basis for promoting adaptability. Dedicated to trying more efficient deep reinforcement learning (DRL) algorithms, this paper proposed a deep q-network (DQN)-based energy and emission management strategy (E&EMS) at first. Then, two distributed DRL algorithms, namely asynchronous advantage actor-critic (A3C) and distributed proximal policy optimization (DPPO), were adopted to propose EMSs, respectively. Finally, emission optimization was taken into account and then distributed DRL-based E&EMSs were proposed. Regarding dynamic programming (DP) as the optimal benchmark, simulation results show that three DRL-based control strategies can achieve near-optimal fuel economy and outstanding computational efficiency, and compared with DQN, two distributed DRL algorithms have improved the learning efficiency by four times.

关键词： Energy management Optimization Hybrid electric vehicles Heuristic algorithms real-time systems Engines Training Asynchronous advantage actor-critic deep reinforcement learning distributed proximal policy optimization energy management strategy hybrid electric vehicle

来源：评论

学校读者我要写书评

暂无评论

parallel Electromagnetic Transient Simulation Method for MMC-based SST Based on Diakoptics

Parallel Electromagnetic Transient Simulation Method for MMC...

引用

IEEE International Symposium on Power Electronics for distributed Generation systems (PEDG)

作者： Yiming Yang Jin Xu Zirun Li Keyou Wang Guojie Li Pan Wu Department of Electrical Engineering Shanghai Jiao Tong University Shanghai China

With the rapid development of renewable energy generation and the distribution network, the dc network has become a hotspot in research and engineering. In the dc network, the power electronic transformer (PET, also known as solid-state transformer, SST) is the crucial interface device. To achieve better characteristics, the SST based on modular-multilevel-converter (MMC) is proposed. The cascade modulization and high switch efficiency of MMC-SST bring challenges to its electromagnetic (EMT) simulation. The numerous submodules increase the order of the node admittance matrix, requiring more computation resources; the high switch frequency requires a smaller time step, decreasing the calculation speed. To deal with these problems, this paper proposes a parallel EMT simulation method based on the diakoptics. This method achieves parallelized calculation at the system level and submodule level by tearing the whole system into subsystems and applying the compact EMTP method. The proposed method is verified in MATLAB and an FPGA-based real-time simulation platform.

关键词：

来源：评论

学校读者我要写书评

暂无评论

All Federated or Not: Optimizing Personal Model Performance in Cross-silo Federated Learning

All Federated or Not: Optimizing Personal Model Performance ...

引用

International Conference on parallel and distributed systems (ICPADS)

作者： Juan Li Yanmin Zhu Jie Wu Weifan Wu Tianzi Zang Liu Lu Nanjing University of Aeronautics and Astronautics China Shanghai Jiao Tong University China Temple University USA

ISBN: (数字)9798331515966

ISBN: (纸本)9798331515973

In cross-silo federated learning (FL), organizations cooperatively train a global model with their local data. The organizations, however, may be heterogeneous in terms of data distributions. In such cases, FL might produce a biased global model that is not optimal for each organization. Then each organization faces several fundamental questions: should I join FL or just remain alone? If joining FL, which organizations should I cooperate with? In this work, we formulate a coalition formation game in cross-silo FL to help organizations choose proper cooperators. We first build an estimation method to predict personal model performance for each organization before FL starts, and we treat performance improvement as individual utility. With estimated utilities, we design a distributed coalition formation algorithm to find stable coalition structures and optimize social welfare at the same time. Our simulations based on MNIST and FMNIST datasets show that the estimation model can predict the sign of the utility correctly with a probability of 0.9 and has an average relative error of $30 \%$. With the above errors, the obtained coalition structure performs well from both perspectives of real social welfare and individual satisfaction.

关键词： Federated learning Neural networks Estimation Organizations Games Predictive models Prediction algorithms Data models distributed algorithms Faces

来源：评论

学校读者我要写书评

暂无评论

Research on High-Dimensional Matrix Multiplication in Machine Learning for MEC

Research on High-Dimensional Matrix Multiplication in Machin...

引用

International Conference on Computer and Communications (ICCC)

作者： Fan Ke Li Chen School of Information Science and Technology University of Science and Technology of China Hefei China

ISBN: (数字)9798331507077

ISBN: (纸本)9798331507084

Machine learning (ML) algorithms have been widely deployed in mobile edge computing (MEC). Howerver, algorithms in MEC always demands high computational power to process complex models and large datasets in real-time. A critical component of this computational workload is matrix multiplication, a fundamental operation in many ML algorithms, particularly in deep learning. To meet the stringent latency and performance requirements of MEC applications, leveraging Graphics Processing Unit (GPU) acceleration for matrix multiplication is essential. GPUs, with their massive parallel processing capabilities, can significantly enhance the efficiency and speed of matrix computations, enabling rapid inference and training at the edge. This acceleration not only reduces the latency but also offloads the central cloud, providing a more distributed and scalable solution for real-time ML applications at the edge. To this end, this paper first introduces the background of using matrix multiplication to accelerate model training speed in ML for MEC, clarifies the significance of accelerating matrix multiplication to ML and MEC, then reviews and experiments on existing schemes for accelerating matrix multiplication and compares and summarizes various schemes from the perspective of different sparsity of matrices on the GPU. Finally, we points out the possible future research directions and technical challenges, and provides the certain reference value for the research in related fields.

关键词： Training Deep learning Machine learning algorithms Multi-access edge computing Reviews Graphics processing units parallel processing real-time systems Inference algorithms Sparse matrices

来源：评论

学校读者我要写书评

暂无评论

When Less is More: Achieving Faster Convergence in distributed Edge Machine Learning

arXiv

引用

arXiv 2024年

作者： Basani, Advik Raj Vivek, Siddharth Chaitra Krishna, Advaith Paul, Arnab K. DaSHLab Department of Computer Science and Information Systems BITS Pilani K. K. Birla Goa Campus India

distributed Machine Learning (DML) on resource-constrained edge devices holds immense potential for real-world applications. However, achieving fast convergence in DML in these heterogeneous environments remains a significant challenge. Traditional frameworks like Bulk Synchronous parallel (BSP) and Asynchronous Stochastic parallel (ASP) rely on frequent, small updates that incur substantial communication overhead and hinder convergence speed. Furthermore, these frameworks often employ static dataset sizes, neglecting the heterogeneity of edge devices and potentially leading to straggler nodes that slow down the entire training process. The straggler nodes, i.e., edge devices that take significantly longer to process their assigned data chunk – hinder the overall training speed. To address these limitations, this paper proposes Hermes, a novel probabilistic framework for efficient DML on edge devices. This framework leverages a dynamic threshold based on recent test loss behavior to identify statistically significant improvements in the model’s generalization capability, hence transmitting updates only when major improvements are detected, thereby significantly reducing communication overhead. Additionally, Hermes employs dynamic dataset allocation to optimize resource utilization and prevents performance degradation caused by straggler nodes. Our evaluations on a real-world heterogeneous resource-constrained environment demonstrate that Hermes achieves faster convergence compared to state-of-the-art methods, resulting in a remarkable 13.22x reduction in training time and a 62.1% decrease in communication overhead. Copyright © 2024, The Authors. All rights reserved.

关键词： Stochastic systems

来源：评论

学校读者我要写书评

暂无评论

distributed Energy Storage SOC Balancing Strategy Based on Adaptive Droop Control with Dynamic Correction Coefficients

Distributed Energy Storage SOC Balancing Strategy Based on A...

引用

Mechatronics Technology and Intelligent Manufacturing (ICMTIM), International Conference on

作者： Xiaoli Qi Houlei Gao Fang Peng Rui Luo Wanshuo Ma School of Electrical Engineering Shandong University Jinan China Jinan City Power Supply Company State Grid Shandong Electric Power Company Jinan China

ISBN: (数字)9798350363265

ISBN: (纸本)9798350363272

Inconsistent State of Charge (SOC) of parallel distributed Energy Storage (DES) can cause issues in microgrid stability and energy storage battery lifespan when using conventional Droop with fixed droop coefficients. DES with low SOC value may discharge prematurely and exit from operation, disrupting the microgrid's stable operation. To address this, this paper proposes an adaptive Droop control strategy based on dynamic correction coefficients that consider the local SOC of each DES and coordinate their output, to maintain stable microgrid operation. Its efficacy is confirmed through simulations on the PSCAD/EMTDC platform.

关键词： Photovoltaic systems Adaptation models Mechatronics Microgrids Stability analysis real-time systems Batteries

来源：评论

学校读者我要写书评

暂无评论

A sensor fusion strategy based on a distributed optical sensing of airframe deformation applied to actuator load estimation 8

A sensor fusion strategy based on a distributed optical sens...

引用

IEEE 8th International workshop on Metrology for AeroSpace (MetroAeroSpace)

作者： Quattrocchi, Gaetano Dalla Vedova, Matteo D. L. Frediani, Emanuele Maggiore, Paolo Berri, Pier Carlo Politecn Torino Dept Mech & Aerosp Engn Turin Italy

ISBN: (纸本)9781728175560

real-time health monitoring of mechatronic onboard systems often involves model-based approaches comparing measured (physical) signals with numerical models or statistical data. This approach often requires the accurate measurement of specific physical quantities characterizing the state of the real system, the command inputs, and the various boundary conditions that can act as sources of disturbance. In this regard, the authors study sensor fusion techniques capable of integrating the information provided by a network of optical sensors based on Bragg gratings to reconstruct the signals acquired by one or more virtual sensors (separately or simultaneously). With an appropriate Fiber Bragg Gratings (FBGs) network, it is possible to measure directly (locally) several physical quantities (e.g. temperature, vibration, deformation, humidity, etc.), and, at the same time, use these data to estimate other effects that significantly influence the system behavior but which, for various reasons, are not directly measurable. In this case, such signals could be "virtually measured" by suitably designed and trained artificial neural networks (ANNs). The authors propose a specific sensing technology based on FBGs, combining suitable accuracy levels with minimal invasiveness, low complexity, and robustness to EM disturbances and harsh environmental conditions. The test case considered to illustrate the proposed methodology refers to a servomechanical application designed to monitor the health status in real-time of the flight control actuators using a model-based approach. Since the external aerodynamic loads acting on the system influence the operation of most of the actuators, their measurement would be helpful to accurately simulate the monitoring model's dynamic response. Therefore, the authors evaluate the proposed sensor fusion strategy effectiveness by using a distributed sensing of the airframe strain to infer the aerodynamic loads acting on the flight control actuator. O

关键词： ANN distributed optical sensing Fiber Bragg Gratings FBG load estimation on-board fault detection optical fibers prognostics smart sensor

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：