Genetic Algorithm-Driven Joint Optimization of Task Offloading and Resource Allocation for Fairness-Aware Latency Minimization in Mobile Edge Computing

Genetic Algorithm-Driven Joint Optimization of Task Offloading and Resource Allocation for Fairness-Aware Latency Minimization in Mobile Edge Computing

Mobile Edge Computing (MEC) alleviates latency and bandwidth strain on centralized cloud infrastructures by enabling the offloading of tasks to proximal edge servers, yet resource optimization in dense dynamic networks remains an open problem. This paper proposes a genetic algorithm (GA)-based appro...

Full description

Saved in:
Bibliographic Details
Main Authors: Mohamed Elkawkagy, Ibrahim A. Elgendy, Samia Allaoua Chelloug, Heba Elbeh
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Edge computing
offloading strategy
genetic algorithm
resource allocation
Online Access:https://ieeexplore.ieee.org/document/11062637/
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Mobile Edge Computing (MEC) alleviates latency and bandwidth strain on centralized cloud infrastructures by enabling the offloading of tasks to proximal edge servers, yet resource optimization in dense dynamic networks remains an open problem. This paper proposes a genetic algorithm (GA)-based approach to jointly optimize three important parameters: 1) the proportion of tasks offloaded to mobile edge servers (MES), 2) channel bandwidth allocation, and 3) computational resource allocation, in order to minimize the total task completion time with fairness among user devices (UDs). Resolving constraints such as limited wireless transmission capacity that is bounded and MES processing resources, the GA effectively explores solution spaces through selection, crossover, and mutation operations without falling into local optima. Simulation outcomes demonstrate 50% reduction in completion time compared to non-offloading strategies, with fairness indicated by standard deviation metrics showing equal performance across UDs. The algorithm scales, with task durations leveling off after 50 generations for 15 UDs and robustness being maintained up to 300 iterations. Results identify important thresholds (e.g., bandwidth > 5 MHz yields diminishing returns) and achieve near-optimal efficiency at high levels of resources. This work advances MEC efficiency by tackling multi-resource optimization issues through metaheuristic exploration, with a scalable approach to latency-sensitive IoT and 5G network use cases.
ISSN:2169-3536

Similar Items