Dynamic resource allocation algorithm of virtual networks in edge computing networks

The deployment and allocation of network resources are important in the application of edge computing. As an important resource allocation technology in edge computing, network virtualization faces the challenge of the virtual network mapping problem. Most existing studies are limited to static reso...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Personal and ubiquitous computing Ročník 25; číslo 3; s. 571 - 586
Hlavní autoři: Xiao, Xiancui, Zheng, Xiangwei, Jie, Tian
Médium: Journal Article
Jazyk:angličtina
Vydáno: London Springer London 01.06.2021
Springer Nature B.V
Témata:
Algorithms
Computer Science
Computer simulation
Design optimization
Edge computing
Error reduction
Mapping
Mathematical analysis
Mobile Computing
Original Article
Personal Computing
Radial basis function
Resource allocation
Training
User Interfaces and Human Computer Interaction
Virtual networks
Dynamic resource allocation
Radial basis function network
Group search optimizer
Virtual network
Edge computing
ISSN:1617-4909, 1617-4917
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:The deployment and allocation of network resources are important in the application of edge computing. As an important resource allocation technology in edge computing, network virtualization faces the challenge of the virtual network mapping problem. Most existing studies are limited to static resource allocation, ignoring the time-varying properties of user resource demands, which results in wasted resources. Since user resource demands vary over time, resource allocation with predictive mechanism is a promising solution. However, there are few studies on the application of predictive algorithm as radial basis function network (RBF) algorithms in virtual network dynamic resource allocation. In addition, due to the excessive use of hidden RBF units, this method suffers from expensive inner product calculations and long training times. In this paper, we propose a dynamic network resource demand predicting algorithm based on the group search optimizer (GSO) and incremental design of the RBF (GSO-INC-RBFDM). In the network mapping, the GSO is first used to optimize the node solution. Then, the incremental design is utilized to eliminate the maximum error value and reduce the inner product calculation and training time by adding the RBF unit one by one. Finally, we apply the improved RBF to predict the user demand and reallocate resources based on the predicted results. Simulation results shows that the GSO-INC-RBFDM demonstrates good performance in terms of the acceptance rate, network cost, link pressure and average revenue compared with traditional algorithms.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1617-4909
1617-4917
DOI:10.1007/s00779-019-01277-2