An Efficient Dynamic Resource Allocation in Network Function Virtualization using An Integrated Machine Learning Model

Authors

  • Jeyakumar D, Rajabhushanam C

Abstract

Network function virtualization is used to eliminate the physical middleware by flexible virtual network functions (VNF). For predicting and balancing the network load for evergreen network traffic, the VNFs need to be instantiated with resource allocation based on the load. Understanding the demands and determining the number of resources is a crucial task. Several earlier methods have allocated a fixed number of resources to each VNF, making waste of resources or not allocating enough resources to fulfil the demands. For solving the above-said problem, this paper proposed an integrated Convolutional Neural Network model for determining the number of resources for each VNF according to their traffic load efficiently. The CNN model integrated with LSTM model is trained with a real VNF dataset, and the performance is verified, whereas the trained model can easily predict the number of resources required to balance the traffic load. The proposed CNN-LSTM model is evaluated using real-time data and finds out how it eliminated under-and-over resource allocation with better quality of service in terms of decreased delay and optimal resource allocation.

Published

2021-09-15

How to Cite

Jeyakumar D, Rajabhushanam C. (2021). An Efficient Dynamic Resource Allocation in Network Function Virtualization using An Integrated Machine Learning Model. Drugs and Cell Therapies in Hematology, 10(1), 2306–2316. Retrieved from http://www.dcth.org/index.php/journal/article/view/433

Issue

Section

Articles