MOSO: multi-objective snake optimizer with density estimation and grid indexing mechanism for edge computing task offloading and scheduling optimization

The proposal of task offloading and scheduling optimization in edge computing aims to effectively alleviate network congestion caused by a large amount of data in ultra-dense networks (UDN), minimize energy consumption during data transmission, reduce transmission latency, and enhance Quality of Ser...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Cluster computing Ročník 28; číslo 4; s. 244
Hlavní autoři: Zhang, Shi-Hui, Wang, Jie-Sheng, Zhang, Si-Wen, Xing, Yu-Xuan, Wang, Xiao-Tian, Sui, Xiao-Fei
Médium: Journal Article
Jazyk:angličtina
Vydáno: New York Springer US 01.08.2025
Springer Nature B.V
Témata:
Algorithms
Cloud computing
Computation offloading
Computer Communication Networks
Computer Science
Convexity
Costs
Data transmission
Decision making
Deep learning
Density
Dynamic programming
Edge computing
Energy consumption
Genetic algorithms
Indexing
Mixed integer
Multiple objective analysis
Network latency
Nonlinear programming
Operating Systems
Optimization
Pareto optimization
Processor Architectures
Resource scheduling
Scheduling
Task scheduling
User experience
Internet of things
Snake optimizer
Multi-objective optimization
Scheduling optimization
Task uninstallation
Edge computing
ISSN:1386-7857, 1573-7543
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:The proposal of task offloading and scheduling optimization in edge computing aims to effectively alleviate network congestion caused by a large amount of data in ultra-dense networks (UDN), minimize energy consumption during data transmission, reduce transmission latency, and enhance Quality of Service (QoS). To address this issue, the Power Allocation (PA) problem for mobile users is presented, aiming to minimize energy consumption through the application of Convex Optimization Techniques. Subsequently, the Joint Request Offloading and Resource Scheduling (JRORS) problem is modeled as a mixed-integer nonlinear programming problem to minimize request response latency and enhance welfare. The JRORS problem can be further divided into two subproblems: the request offloading problem and the computational resource scheduling problem. The optimal solution is sought through mathematical modeling. A multi-objective snake optimizer based on density estimation and grid index mechanism is proposed to solve the multi-objective problem of edge computing task offloading and scheduling optimization. Firstly, applying the grid indexing mechanism transforms the single-objective snake optimizer to a multi-objective algorithm. Secondly, a roulette wheel selection method is utilized to choose the elite solutions, referred to as "Leaders". Density estimation is conducted based on the dominance capability of individuals, employing fast non-dominated sorting and density estimation for dual optimization of the archive. This approach not only yields an optimal Pareto solution set but also excludes similar individuals before the next iteration, ensuring population diversity and enhancing the effectiveness of the multi-objective snake optimizer. The proposed improved algorithm demonstrates outstanding performance across various test functions and conducts multi-objective optimization of energy consumption and efficiency for task offloading under nine different user scales. Comparative results against NSGA-II, MOPSO, MSSA, and MOALO show that MOSO effectively finds representative samples that achieve the optimal balance between energy consumption and welfare. The results further show that MOSO can maintain good performance in ultra-dense edge computing (UDEC) networks, find a set of representative samples with a wider range and better frontier between energy consumption and welfare, and complete the joint optimization between thm.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1386-7857
1573-7543
DOI:10.1007/s10586-024-04902-9