Research On Optical Network Virtualization Resource Scheduling Technology

With the rapid development of the Internet,the Internet architecture faces new demands and challenges due to the community of Internet user continues to grow and network application traffic grows exponentially.Faced with the demand for new Internet diversity services,the problems of poor flexibility and insufficient reliability of the traditional Internet and network rigidity has become increasingly prominent.Optical network virtualization can overcome some of the problems faced by the traditional Internet.In an optical network virtualization environment,multiple virtual networks are mapped to a common physical network to share resources of the underlying infrastructure.One of the challenges of network virtualization is how to map multiple virtual networks with resource constraints to physical networks in order to effectively utilize the underlying physical resources,which has become a research hotspot of network virtualization.This paper takes the mapping algorithm as the research object in the optical network virtualization environment,and conducts in-depth research from several aspects such as resource management architecture,dynamic resource allocation algorithm and energy-saving mapping algorithm.The main research contents of this paper are as follows:Above all,this paper describes the resource management architecture according to the characteristics of virtual resources,analyzes the characteristics and functions of the resource management architecturen and presents a resource management architecture based on multi-domain and hierarchy suitable for multiple domains.The global scheduling center is mainly responsible for accepting virtual network requests and subnetting,and each local scheduling center is responsible for completing resource mapping of virtual subnets.The hierarchical and domain-based resource management architecture improves the scalability and autonomy and reduces the complexity of resource management.Secondly,aiming at the problems of large resource overhead and low efficiency of traditional virtual network mapping algorithm,a heuristic dynamic resource allocation algorithm is proposed.In this paper,the dynamic resource allocation algorithm is divided into active and passive cases.Active resource allocation refers to the infrastructure provider actively adjusting the physical network according to the current physical link and node load.Passive resource allocation refers to resource redistribution by infrastructure providers based on resource requirements requested by virtual networks.This paper describes the above two problems as a mixed integer linear programming model(MILP)and model the reconfiguration cost.In the dynamic resource allocation algorithm DVNMA_NEW proposed in this paper,how to minimize the cost to achieve resource reconfiguration is studied while satisfying the resource requirements of the virtual network.The simulation results show that the proposed dynamic resource allocation algorithm reduces the resource overhead cost of the network during the reconfiguration process,increases the acceptance rate of virtual network requests and improves resource utilization.Thirdly,an efficient heuristic energy-saving mapping algorithm is proposed in order to solve the problem that low resource and energy utilization caused by the current network is designed for peak load and over-provision.The algorithm aims to minimize the energy consumption of the underlying infrastructure and the virtual network energy-saving mapping described a MILP problem.In the heuristic energy-saving mapping algorithm PEVNP_NEW designed in this paper,two problems of power consumption and load balancing are considered simultaneously for takes into account the two performance indicators of power consumption and acceptance rate.In the node mapping,this paper adopts the principle of minimum voltage priority to achieve the purpose of load balancing.In the link mapping,the weighted shortest path method is adopted to avoid the bottleneck link and the overheated node.The simulation results show that the algorithm effectively reduces the underlying basis of the power consumption of the facility and improves the acceptance rate of virtual requests.