Effective and efficient AI-based approaches to cloud resource provisioning

This paper aims to design efficient and effective approaches to virtual network embedding (VNE) problem, which deals with the embedding of a requested virtual network (VN) in an underlying physical (substrate network) infrastructure. When the node and link constraints (including CPU, memory, network bandwidth, and network delay) are both taken into account, the VN embedding problem is NP-hard, even in the offline case. The capabilities of some Artificial Intelligence (AI) techniques have been validated in handling the VN problem. In this paper, we propose two efficient and effective VNE algorithms based on differential evolutionary (DE) technique. The extensive simulation results show that DE technique performs some orders of magnitude faster than GA and PSO-based VNE algorithms in achieving the comparable long-term revenue of Infrastructure providers.