AI-based Resource Prediction in Network Function Virtualization Architectures

Abstract

The high reconfiguration time of virtualised networks led to the definition of allocation procedures based on the prediction of the processing resources required. We propose an Artificial Intelligence-based resource allocation procedure in which the use of processing resources is monitored and the resources to be allocated are accordingly predicted. We evaluate the impact on the costs of the proposed allocation procedure and show that the cost increase is limited with respect to the case of exact knowledge of the needed processing resources.