An Energy Resource Management for Cluster based IoHV Supported by Fog Computing DOI Creative Commons
Ahmed Jawad Kadhim

Karbala International Journal of Modern Science, Journal Year: 2024, Volume and Issue: 11(1)

Published: Dec. 14, 2024

Internet of Hybrid Vehicle Networks (IoHV) is a network generated by merging the with Vehicular Ad-Hoc Network (H-VANET). In IoHV, various types electric and fuel vehicles create tasks. However, executing several tasks affects their lifetime because they suffer from energy limitation issues which one IoHV challenges. On other hand, fog nodes have unlimited can be used to execute most quickly. this paper, we produce new Energy Resource management Technique for called ERTH that aims offload nodes. The main goal saving vehicles. As result, probability switching off number, time, cost recharging batteries will reduced. Moreover, propose energy-aware clustering method connect vehicles, It help save balance load. results showed better than PLIFS RMOIE according EC. NDEV approximately 47.2% 42.4% different numbers 26.19% 14.1% mobility speeds less RMOIE, respectively. Finally, PET 2.1% 3.09% 1.9% 3.18% more

Language: Английский

Machine learning-based computation offloading in edge and fog: a systematic review DOI

Sanaz Taheri-abed,

Amir Masoud Eftekhari Moghadam, Mohammad Hossein Rezvani

et al.

Cluster Computing, Journal Year: 2023, Volume and Issue: 26(5), P. 3113 - 3144

Published: July 21, 2023

Language: Английский

Citations

17
Comprehensive survey on resource allocation for edge-computing-enabled metaverse DOI

Tanmay Baidya,

Sangman Moh

Computer Science Review, Journal Year: 2024, Volume and Issue: 54, P. 100680 - 100680

Published: Sept. 9, 2024

Language: Английский

Citations

7
DRL-based dependent task offloading with delay-energy tradeoff in medical image edge computing DOI Creative Commons
Qi Liu, Zhao Tian, Ning Wang

et al.

Complex & Intelligent Systems, Journal Year: 2024, Volume and Issue: 10(3), P. 3283 - 3304

Published: Jan. 29, 2024

Abstract Task offloading solves the problem that computing resources of terminal devices in hospitals are limited by massive radiomics-based medical image diagnosis model (RIDM) tasks to edge servers (ESs). However, sequential decision-making is NP-hard. Representing dependencies and developing collaborative between ESs have become challenges. In addition, model-free deep reinforcement learning (DRL) has poor sample efficiency brittleness hyperparameters. To address these challenges, we propose a distributed dependent task strategy based on DRL (DCDO-DRL). The objective maximize utility RIDM tasks, which weighted sum delay energy consumption generated execution. modeled as directed acyclic graph (DAG). sequence prediction S2S neural network adopted represent decision process within DAG. Next, processing algorithm designed layer further improve run efficiency. Finally, DCDO-DRL follows discrete soft actor-critic method robustness network. numerical results prove convergence statistical superiority strategy. Compared with other algorithms, improves execution at least 23.07, 12.77, 8.51% three scenarios.

Language: Английский

Citations

5
An Auction-Based Bid Prediction Mechanism for Fog-Cloud Offloading Using Q-Learning DOI Creative Commons

Reza Besharati,

Mohammad Hossein Rezvani,

Mohammad Mehdi Gilanian Sadeghi

et al.

Complexity, Journal Year: 2023, Volume and Issue: 2023, P. 1 - 20

Published: Jan. 17, 2023

In the fog computing paradigm, if resources of an end device are insufficient, user’s tasks can be offloaded to nearby devices or central cloud. addition, due limited energy mobile devices, optimal offloading is crucial. The method presented in this paper based on auction theory, which has been used recent studies optimize computation offloading. We propose a bid prediction mechanism using Q-learning. Nodes participating announce value auctioneer entity, and node with highest winner. Then, only winning right offload its upstream (parent) node. main idea behind Q-learning that it stateless considers current state perform action. evaluation results show values predicted by near-optimal. On average, proposed consumes less than traditional state-of-the-art techniques. Also, reduces execution time leads consumption network resources.

Language: Английский

Citations

11
EDITORS: Energy-aware Dynamic Task Offloading using Deep Reinforcement Transfer Learning in SDN-enabled Edge Nodes DOI Creative Commons
Thar Baker, Zaher Al Aghbari, Ahmed M. Khedr

et al.

Internet of Things, Journal Year: 2024, Volume and Issue: 25, P. 101118 - 101118

Published: Feb. 10, 2024

In mobile edge computing systems, a task offloading approach should balance efficiency, adaptability, trust management, and reliability. This aims to maximize resource utilization, improve user experience, satisfy application-specific requirements while taking into account the dynamic limited nature of environments. Additionally, tasks, these systems are vulnerable several attacks privacy breaches, necessitating node evaluation. However, not all necessary features present in methods currently used. research proposes 'EDITORS' (Energy-efficient DynamIc Task Offloading method utilising Deep Reinforcement Transfer Learning (DRTL) Software-Defined Network (SDN) enabled environments), novel aimed at addressing multifaceted issues associated with systems. The primary goal EDITORS is design system that incorporates trusted nodes prioritizing energy timeliness, reliability, outperforming existing terms quality plan. uses DRTL agents nodes, which communicate SDN controller learn most appropriate choices based on network conditions availability. Extensive simulations (six) conducted show significantly increases efficiency preserving low-latency completion compared five (DDLO, DROO, DMRO, DRL without TL SDN). includes evaluation, device prediction using LSTM, adaptation newly added devices through transfer learning, unlike other just concentrate offloading.

Language: Английский

Citations

4
Delay-energy-aware joint multi-cell association, service caching, and task offloading in hybrid-task heterogeneous edge computing networks DOI
Bassant Tolba, Maha Elsabrouty, Mohammed Abo‐Zahhad

et al.

Computer Networks, Journal Year: 2025, Volume and Issue: unknown, P. 111231 - 111231

Published: March 1, 2025

Language: Английский

Citations

0
A tripartite evolutionary game for strategic decision-making in live-streaming e-commerce DOI Creative Commons
Georgia Fargetta, Laura Scrimali

Journal of Computational Science, Journal Year: 2025, Volume and Issue: unknown, P. 102585 - 102585

Published: April 1, 2025

Language: Английский

Citations

0
Promoting Data Sharing Diffusion in the Automotive Industry: An Evolutionary Game Model on Complex Networks DOI
Xiaochuan Tang,

Tao Lan,

Fan Du

et al.

Expert Systems with Applications, Journal Year: 2025, Volume and Issue: unknown, P. 127992 - 127992

Published: May 1, 2025

Language: Английский

Citations

0
A collaborative operation mode of energy storage system and train operation system in power supply network DOI
Songpo Yang, Yanyan Chen,

Zhurong Dong

et al.

Energy, Journal Year: 2023, Volume and Issue: 276, P. 127617 - 127617

Published: April 21, 2023

Language: Английский

Citations

9
Solving the 0/1 Knapsack Problem Using Metaheuristic and Neural Networks for the Virtual Machine Placement Process in Cloud Computing Environment DOI Creative Commons
Mohamed Abid, Said El Kafhali, Abdellah Amzil

et al.

Mathematical Problems in Engineering, Journal Year: 2023, Volume and Issue: 2023(1)

Published: Jan. 1, 2023

Virtual machine placement (VMP) is carried out during virtual migration to choose the best physical computer host machines. It a crucial task in cloud computing. directly affects data center performance, resource utilization, and power consumption, it can help providers save money on maintenance. To optimize various characteristics that affect centers, VMs, their runs, numerous VMP strategies have been developed computing environment. This paper aims compare accuracy efficiency of nine distinct for treating as knapsack problem. In numerical analysis, we test conditions determine how well system works. We first illustrate rate convergence algorithms, then execution time growth given number machines, lastly development CPU usage supplied by methods throughout three analyzed conditions. The obtained results reveal neural network algorithm performs better than other eight approaches. model performed well, shown its ability provide near‐optimal solutions cases.

Language: Английский

Citations

7