Virtualized Network Function Scaling Strategy Based on Additional Momentum BP Neural Network

• DOI: 10.1145/3640912.3640922
• Publication Date: 2024-02-22T18:37:41Z
• Authors: Ji Qi; Yuehui Tan; Qifan Zhao; Weiliang Li; Bao Sun; Yi Wu; Hui Tao; Xin Zhang
• Keywords: Momentum (technical analysis)

With the increasing demand for data, mobile operator face challenge of resource reuse and cost reduction, while also needing to improve network performance support dynamic scaling. In this paper, we propose a Virtualized Network Function (VNF) scaling ratio prediction model based on an additional momentum BP neural optimize VNF instances in context public cloud Virtualization (NFV) scenarios. We monitoring data using method, resulting specific strategies. The results show that is highly effective predicting number instances, with error rate only 0.0009 accuracy 98%. Compared unimproved model, method leads smaller errors notable improvements both convergence speed accuracy.