Virtual Network Function Resource Requirements Prediction Model Based on CNN-GRU

Chunqiao Mao, Peng Yi, Dan Li, Juan Shen

Abstract: The rise of network function virtualization (NFV) technology makes the realization of network function change from hardware middleware to virtual network function (VNF). The existing methods allocate fixed resources to each VNF instance, but this resource allocation method will cause resource waste, which will affect the quality of service. The resource requirements prediction model solves the resource allocation problem by predicting the change of resource requirements. In this paper, deep learning is used to solve the regression prediction problem, and a VNF resource requirements prediction model based on convolutional neural network (CNN) and gated recurrent unit (GRU) is proposed. Compared with other single model and combined model, the experimental results show that the prediction error rate of the proposed model is reduced by 23.3%.