Research on Resource Reservation Strategy for Edge Federation

With the increasing number of users and the continuous expansion of task scales, the resource constraints faced by edge computing are becoming increasingly pronounced. To address these challenges, edge computing has gradually evolved into an edge federation computing model. This model enhances overa...

Full description

Saved in:
Bibliographic Details
Main Authors: Hengzhou Ye, Huangran Li, Jiaming Li, Qiu Lu, Gong Chen
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/11078238/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849714581771911168
author Hengzhou Ye
Huangran Li
Jiaming Li
Qiu Lu
Gong Chen
author_facet Hengzhou Ye
Huangran Li
Jiaming Li
Qiu Lu
Gong Chen
author_sort Hengzhou Ye
collection DOAJ
description With the increasing number of users and the continuous expansion of task scales, the resource constraints faced by edge computing are becoming increasingly pronounced. To address these challenges, edge computing has gradually evolved into an edge federation computing model. This model enhances overall resource utilization through multi-node resource sharing within the federation, effectively alleviating the issue of insufficient resources at individual edge nodes. This paper first analyzes the necessity of implementing a resource reservation strategy within the edge federation. It proposes a load prediction-based resource reservation strategy and optimizes this strategy according to the load levels present in the edge federation. Subsequently, we introduce a multi-agent deep deterministic policy gradient (RRP-MADDPG) approach based on a multi-agent deep reinforcement learning algorithm aimed at reducing average task delay. Simulation results demonstrate that both proposed resource reservation strategies can significantly reduce average task delays. Furthermore, the RRP-MADDPG strategy exhibits excellent convergence performance and outperforms both the load prediction-based reservation strategy and other similar deep reinforcement learning algorithms.
format Article
id doaj-art-a457eeeef89f4fcb8a29f131ee21c5f8
institution DOAJ
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-a457eeeef89f4fcb8a29f131ee21c5f82025-08-20T03:13:39ZengIEEEIEEE Access2169-35362025-01-011312440212441210.1109/ACCESS.2025.358822411078238Research on Resource Reservation Strategy for Edge FederationHengzhou Ye0https://orcid.org/0000-0001-6646-8747Huangran Li1Jiaming Li2Qiu Lu3Gong Chen4https://orcid.org/0009-0004-6071-8548School of Computer Science and Engineering, Guilin University of Technology, Guilin, ChinaSchool of Computer Science and Engineering, Guilin University of Technology, Guilin, ChinaSchool of Computer Science and Engineering, Guilin University of Technology, Guilin, ChinaSchool of Computer Science and Engineering, Guilin University of Technology, Guilin, ChinaSchool of Computer Science and Engineering, Guilin University of Technology, Guilin, ChinaWith the increasing number of users and the continuous expansion of task scales, the resource constraints faced by edge computing are becoming increasingly pronounced. To address these challenges, edge computing has gradually evolved into an edge federation computing model. This model enhances overall resource utilization through multi-node resource sharing within the federation, effectively alleviating the issue of insufficient resources at individual edge nodes. This paper first analyzes the necessity of implementing a resource reservation strategy within the edge federation. It proposes a load prediction-based resource reservation strategy and optimizes this strategy according to the load levels present in the edge federation. Subsequently, we introduce a multi-agent deep deterministic policy gradient (RRP-MADDPG) approach based on a multi-agent deep reinforcement learning algorithm aimed at reducing average task delay. Simulation results demonstrate that both proposed resource reservation strategies can significantly reduce average task delays. Furthermore, the RRP-MADDPG strategy exhibits excellent convergence performance and outperforms both the load prediction-based reservation strategy and other similar deep reinforcement learning algorithms.https://ieeexplore.ieee.org/document/11078238/Edge federationtask schedulingresource reservationmulti-agent
spellingShingle Hengzhou Ye
Huangran Li
Jiaming Li
Qiu Lu
Gong Chen
Research on Resource Reservation Strategy for Edge Federation
IEEE Access
Edge federation
task scheduling
resource reservation
multi-agent
title Research on Resource Reservation Strategy for Edge Federation
title_full Research on Resource Reservation Strategy for Edge Federation
title_fullStr Research on Resource Reservation Strategy for Edge Federation
title_full_unstemmed Research on Resource Reservation Strategy for Edge Federation
title_short Research on Resource Reservation Strategy for Edge Federation
title_sort research on resource reservation strategy for edge federation
topic Edge federation
task scheduling
resource reservation
multi-agent
url https://ieeexplore.ieee.org/document/11078238/
work_keys_str_mv AT hengzhouye researchonresourcereservationstrategyforedgefederation
AT huangranli researchonresourcereservationstrategyforedgefederation
AT jiamingli researchonresourcereservationstrategyforedgefederation
AT qiulu researchonresourcereservationstrategyforedgefederation
AT gongchen researchonresourcereservationstrategyforedgefederation