AI-based survivable design for hybrid virtual networks for single regional failures in cloud data centers

Network virtualization is a key technology that enables a substrate infrastructure to be shared by multiple virtual heterogeneous networks (VNs). Recent research of network virtualization typically focuses on designing a variety of methods to perform the mapping between the virtual and physical networks. Nevertheless, these related approaches and algorithms could either have good performance in building unicast service-oriented virtual networks or be applied to virtual multicast service-oriented networks. A limited number of studies have focused on the mapping problem in survivable hybrid virtual networks (HVNs), which is critical in considering both the multicast and unicast traffic of the same VN request. In this paper, we research the survivable HVN mapping problem while considering regional failures of the substrate network, and we posit an AI-based high-efficiency framework and algorithm to resolve this problem. Groups of simulations are conducted under different scenarios and compared with the existing approach to evaluate the framework and algorithms proposed in this paper. The simulation results demonstrate that the performance of our approach is superior to that of the existing approach.