检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Guliyev, Rustam Haldar, Aparajita Ferhatosmanoglu, Hakan University of Warwick Coventry United Kingdom Amazon Web Services United States

Graph Neural Network (GNN) models on streaming graphs entail algorithmic challenges to continuously capture its dynamic state, as well as systems challenges to optimize latency, memory, and throughput during both inference and training. We present D3-GNN, the first distributed, hybrid-parallel, streaming GNN system designed to handle real-time graph updates under online query setting. Our system addresses data management, algorithmic, and systems challenges, enabling continuous capturing of the dynamic state of the graph and updating node representations with fault-tolerance and optimal latency, load-balance, and throughput. D3-GNN utilizes streaming GNN aggregators and an unrolled, distributed computation graph architecture to handle cascading graph updates. To counteract data skew and neighborhood explosion issues, we introduce inter-layer and intra-layer windowed forward pass solutions. Experiments on large-scale graph streams demonstrate that D3-GNN achieves high efficiency and scalability. Compared to DGL, D3-GNN achieves a significant throughput improvement of about 76x for streaming workloads. The windowed enhancement further reduces running times by around 10x and message volumes by up to 15x at higher parallelism. Copyright © 2024, The Authors. All rights reserved.

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

distributed Tube-Based Nonlinear MPC for Motion Control of Skid-Steer Robots With Terra-Mechanical Constraints

引用

IEEE ROBOTICS AND AUTOMATION LETTERS 2021年第4期6卷 8045-8052页

作者： Prado, Alvaro J. Torres-Torriti, Miguel Cheein, Fernando A. Pontificia Univ Catolica Chile Sch Engn Dept Elect Engn Robot & Automat Lab Valparaiso 2390144 Chile Pontificia Univ Catolica Chile Dept Elect Engn Valparaiso 2390144 Chile Univ Tecn Federico Santa Maria Dept Elect Engn Valparaiso 5400 Chile

Strategies to reduce slippage and disturbing wheelterrain interactions are essential to improve navigation and motion control of field robots. Thus, this work proposes an integral control architecture based on a distributed tube-based nonlinear Model Predictive Control scheme to regulate tire dynamics and an adaptive model-based control scheme for trajectory tracking over deformable terrain. For the proposed control architecture, the overall system is decomposed into simpler subsystems to separately represent the four-tire driven motion dynamics (i.e., slip and sideslip) from that of the vehicle's pose and speeds. Since a vehicle and its tires have different dynamic response characteristics, cooperative agents of the distributed control strategy are able to exchange information between subsystems to attain evenly allocated drivetrain torques during slippery situations. The motion controller is made adaptive to terra-mechanical parameters with a NonlinearMoving Horizon Estimation approach working under a parallel real-time Iteration scheme. Field experimentations in an industrial compact loader Cat degrees 262 C subject to off-road conditions demonstrated that the proposed approach was capable of reducing up to a minimum of 18.2% of tire slip and sidelip range of +/- 6.6 degrees when compared to its non-robust counterpart. Consequently, the proposed approach was also able to reduce lateral and longitudinal trajectory tracking errors by around 66.6% and 43.7%, respectively, which may have a direct impact on the resources of the machinery.

关键词： Motion control distributed robot systems robust control mining robotics field robots

来源：评论

学校读者我要写书评

暂无评论

Flexible Matrix of Controllers for real time parallel Control

引用

ENERGIES 2022年第5期15卷 1833-1833页

作者： Chaber, Patryk Wojtulewicz, Andrzej Warsaw Univ Technol Fac Elect & Informat Technol Inst Control & Computat Engn Ul Nowowiejska 15-19 PL-00665 Warsaw Poland

This work aims to develop a novel system, including software and hardware, to perform independent control tasks in a genuine parallel manner. Currently, to control processes with various sampling periods, distributed control systems are most commonly utilized. The main goal of this system is to propose an alternative solution, which allows simultaneous control of both fast and slow processes. The presented approach utilizes FPGA (Field Programmable Gate Array) with Nios II processor (Intel Soft Processor Series) to implement and maintain instances of independent controllers. Instances can implement FDMC (Fast Dynamic Matrix Control) and PID (Proportional-Integral-Derivative) control algorithms with various sampling times. The FPGA-based design allows for true independence of controllers' execution both from one another and the managing processor. Also, pure parallel execution allows for implementing slow and fast controllers in the same device. The complete flexible system with a matrix of controllers working in parallel in real-time was tested with both simulated and actual control processes (servomotor), yielding the same results as fully simulated experiments.

关键词： Model Predictive Control Fast Dynamic Matrix Controller field programmable gate array servomotor matrix of controllers

来源：评论

学校读者我要写书评

暂无评论

Proceedings - 2021 20th International Symposium on parallel and distributed Computing, ISPDC 2021

Proceedings - 2021 20th International Symposium on Parallel ...

引用

20th International Symposium on parallel and distributed Computing, ISPDC 2021

ISBN: (纸本)9781665432818

The proceedings contain 17 papers. The topics discussed include: automatic traffic light preemption for intelligent transportation systems;towards an elastic lock-free Hash Trie design;a novel server-side aggregation strategy for federated learning in non-IID situations;an asynchronous distributed-memory optimization solver for two-stage stochastic programming problems;translation based self-reconfiguration algorithm for 6-lattice modular robots;curator - a system for creating data sets for behavioral malware detection;parallel and distributed task-based Kirchhoff seismic pre-stack depth migration application;periodicity detection algorithm and applications on IoT data;parallel cloud movement forecasting based on a modified boids flocking algorithm;and efficient real-time earliest deadline first based scheduling for Apache spark.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Deadline-aware multi-objective IoT services placement optimization in fog environment using parallel FFD-genetic algorithm

引用

PERVASIVE AND MOBILE COMPUTING 2023年 92卷

作者： Saadian, Fatemeh Motameni, Homayun Golsorkhtabaramiri, Mehdi Islamic Azad Univ Dept Comp Engn Babol Branch Babol Iran Islamic Azad Univ Dept Comp Engn Sari Branch Sari Iran

Nowadays, Internet of things has become as an inevitable aspect of humans' IT-based life. A huge number of geo-distributed IoT enabled devices such as smart phones, smart cameras, health care systems, vehicles, etc. are connected to the Internet and manage users' applications. The IoT applications are generally time sensitive, so that giving them up to Cloud and receiving the response may violate their required deadline, due to distance between user device and centralized Cloud data center and consequently increasing network latency. Fog environment, as an intermediate layer between Cloud and IoT devices, brings a smaller scales of Cloud capabilities closer to user location. Processing real time applications in Fog layer helps more deadlines to be met. Although Fog computing enhances quality of service parameters, limited resources and power of Fog nodes is a challenge in processing applications. Furthermore, the network latency is still an issue for communications between applications' services and between user device and Fog node, which seriously threatens deadline condition. Regarding to mentioned points, this paper proposes a 3-partite deadline-aware applications' services placement optimization model in Fog environment which optimizes total power consumption, total resources wastage, and total network latency, simultaneously. The proposed model prioritizes applications in 3 levels based on their associated deadline, and then the model is solved using a parallel model of first fit decreasing and genetic algorithm combination. Simulations results indicates the superiority of proposed approach against counterpart algorithms in terms of reducing power consumption, resource wastage, network latency, and service rejection rate. (c) 2023 Elsevier B.V. All rights reserved.

关键词： Fog computing Internet of things Service Placement Problem real time applications Genetic algorithm

来源：评论

学校读者我要写书评

暂无评论

parallel real-time simulation on commodity hardware with reusable power system models 21

Parallel real-time simulation on commodity hardware with reu...

引用

Proceedings of the 2021 IEEE/ACM 25th International Symposium on distributed Simulation and real time Applications

作者： Raphael Eidenbenz Carsten Franke Mats Larsson Alexandru Moga Thanikesavan Sivanthi Hitachi ABB Power Grids Research Switzerland Munich University of Applied Sciences Germany Hitachi ABB Power Grids Switzerland ABB Corporate Research Switzerland

ISBN: (纸本)9781665433266

This paper proposes a real-time power system simulation framework that is capable of simulating steady state and electromechanical transients of power systems, with sub-millisecond time resolution. The framework can integrate power system component models packed as reusable Functional Mockup Units (FMUs) to flexibly create power system simulations without the need to recreate new models for different power systems. The integration of individual components is based on a novel model decomposition method, which enables the FMU reuse in different system contexts, as well as a parallel simulation execution onto multi-core machines. Furthermore, the paper proposes methods to optimize the allocation of components to cores and shows that the framework can simulate a medium voltage distributed electrical grid of about 20 components in real-time on a commodity multi-core machine.

关键词：

来源：评论

学校读者我要写书评

暂无评论

FogBus2: A lightweight and distributed container-based framework for integration of IoT-enabled systems with edge and cloud computing

FogBus2: A lightweight and distributed container-based frame...

引用

2021 International workshop on Big Data in Emergent distributed Environments, BiDEDE 2021 - In conjunction with the 2021 ACM SIGMOD/PODS Conference

作者： Deng, Qifan Goudarzi, Mohammad Buyya, Rajkumar University of Melbourne Australia

ISBN: (纸本)9781450384650

Edge/Fog computing is a novel computing paradigm that provides resource-limited Internet of Things (IoT) devices with scalable computing and storage resources. Compared to cloud computing, edge/fog servers have fewer resources, but they can be accessed with higher bandwidth and less communication latency. Thus, integrating edge/fog and cloud infrastructures can support the execution of diverse latency-sensitive and computation-intensive IoT applications. Although some frameworks attempt to provide such integration, there are still several challenges to be addressed, such as dynamic scheduling of different IoT applications, scalability mechanisms, multi-platform support, and supporting different interaction models. To overcome these challenges, we propose a lightweight and distributed container-based framework, called FogBus2. It provides a mechanism for scheduling heterogeneous IoT applications and implements several scheduling policies. Also, it proposes an optimized genetic algorithm to obtain fast convergence to well-suited solutions. Besides, it offers a scalability mechanism to ensure efficient responsiveness when either the number of IoT devices increases or the resources become overburdened. Also, the dynamic resource discovery mechanism of FogBus2 assists new entities to quickly join the system. We have also developed two IoT applications, called Conway's Game of Life and Video Optical Character Recognition to demonstrate the effectiveness of FogBus2 for handling real-time and non-real-time IoT applications. Experimental results show FogBus2's scheduling policy improves the response time of IoT applications by 53% compared to other policies. Also, the scalability mechanism can reduce up to 48% of the queuing waiting time compared to frameworks that do not support scalability. © 2021 ACM.

关键词： Internet of things

来源：评论

学校读者我要写书评

暂无评论

A novel spontaneous control for autonomous microgrid VSC system using BPF droop and improved hysteresis band control scheme

引用

ELECTRIC POWER systems RESEARCH 2023年第1期220卷

作者： Behera, Manoja Kumar Saikia, Lalit Chandra Natl Inst Technol Silchar Dept Elect Engn Silchar Assam India

Various researchers have actively addressed several decentralised control techniques for voltage source converter (VSC) dominated islanded microgrid over the years. However, voltage and frequency control, proper power -sharing, and power quality are open issues for autonomous microgrid VSC systems. A unique voltage and fre-quency control approach is proposed for an autonomous microgrid VSC-based distributed generation (DG) system. The proposed scheme has three folds. An improved double band hysteresis current controller (IDBHCC) is proposed for a three-phase inverter interfaced DG in an islanded microgrid application. The IDBHCC scheme encompasses a common-mode third-harmonic to the current control error, which extends the linear operating range of the VSC and also effectively reduces the total harmonic distortion (THD), thus improving the power quality. Additionally, effective current regulation is achieved by separating common-mode interacting current from the measured current. Furthermore, a proportional-integral fractional-order derivative with filter (PIFODN) controller is introduced for the inner loop voltage control, which provides extra flexibility owing to its fractional properties. A sparrow search algorithm is employed to tune the controller parameters for optimum PIFODN controller control action. Bandpass filter droop, an analogous secondary control method with plug-and-play capabilities, is adopted to enable appropriate load sharing among parallel DG inverters and alleviate fre-quency and voltage variation. The proposed control scheme performance is validated via SimPowerSystem tools in the Matlab Simulink for single and multi DG microgrid systems. The proposed control method has also been tested in the OPALRT-4510 real-time simulator as well. The results have shown that the proposed control method is very good at controlling voltage and frequency with a quick response time. The PIFODN controller has a better dynamic response and is better at handling c

关键词： BPF droop distributed generation Improved double-band hysteresis control scheme Fractional-order control Islanded microgrid Sparrow search algorithm

来源：评论

学校读者我要写书评

暂无评论

Efficient Scaling on GPU for Federated Learning in Kubernetes: A Reinforcement Learning Approach

Efficient Scaling on GPU for Federated Learning in Kubernete...

引用

International Technical Conference o Circuits/systems, Computers and Communications (ITC-CSCC)

作者： Charn-Doh Bak Seung-Jae Han Department of Computer Science & Engineering Yonsei University Seoul Korea

ISBN: (数字)9798350379051

ISBN: (纸本)9798350379068

distributed learning enables efficient training with large-scale data, allowing processing to occur in multiple locations without centralization. However, the “straggler problem”, refer-ring to delayed update due to performance heterogeneity, hinders efficient learning. GPUs are crucial for enhancing learning speed by efficiently processing large-scale data through parallel computing capabilities. Efficient GPU utilization, considering resource costs, is essential for reducing training time and costs. Meanwhile, Kubernetes offers scalability, management ease, and resource provisioning for efficient operations. In this paper, we propose a Kubernetes-based vertical scaling scheme to address FL's straggler problem. For GPU scaling, we formulate an opti-mization problem that considers both resource cost and learning speed. Then, we propose a DRL-based approach to address this problem. We also leverage various Kubernetes features as well as CUDA Multi-Process Service (MPS) for execution of vertical scaling. We validate the proposed scheme's performance through various evaluations on a real testbed.

关键词： Training Costs Federated learning Distance learning Scalability Graphics processing units distributed databases

来源：评论

学校读者我要写书评

暂无评论

Learning complex nonlinear dynamics of a 3D translational parallel manipulator using neural network

引用

HELIYON 2024年第18期10卷 e37669页

作者： Elgammal, Abdullah T. Magdy, Mahmoud Lashin, Manar British Univ Egypt El Sherouk City Egypt Benha Univ Benha Fac Engn Elect Engn Dept Banha Egypt Benha Univ Mech Engn Dept Benha Fac Engn Banha Egypt

This study investigates modeling the dynamics of a 3D translational parallel manipulator with closed chains using feedforward neural networks (FFNNs). The dataset exceeds 50,000 samples, incorporating experimental data collected from a robot prototype using MATLAB (R) real-time workshop and the National Instruments (TM) DAQ toolbox, as well as CAD simulation data from MSC ADAMS software. While achieving satisfactory mean squared error (MSE), some predictions did not fully capture the manipulator's dynamics, with small overfitting observed. A Deep Neural Network (DNN) was tested but faced overfitting and high computational costs, despite being trained on a subset of the dataset. This highlighted the limitations of DNNs for modeling such complicated parallel robots with closed chains and parallelograms. FFNNs were preferred for their simplicity and lower overfitting risk. L2 regularization and k-fold validation were applied to improve performance. Transfer learning (TL) was also employed, fine-tuning a new network with weights from pre-trained FFNNs using a smaller, unseen dataset. This approach significantly reduced MSE and completely eliminated overfitting, demonstrating the effectiveness of TL in refining model performance for forward and inverse dynamics. These findings suggest that FFNNs, combined with TL, L2 regularization, and k-fold validation, offer a robust method for accurately modeling complex robotic dynamics, enhancing control and optimization strategies for complicated robotic systems. Training for all networks was conducted within the MATLAB (R) environment.

关键词： parallel robot Feedforward neural networks (FFNNs) Transfer learning Machine learning Forward/inverse dynamics L2 regularization K-fold validation

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：