Research On Virtual Network Mapping For Heterogeneous Software-Defined Networks

With the continuous development and advancement of cloud computing and 5G technology,the scale of the network continues to expand.In order to adapt to the business needs of users,heterogeneous and virtualization have become the development trend of future networks.The heterogeneous network consists of multiple different types of networks,each of which has different network characteristics.Resource allocation and virtualization in this complicated access environment will be very difficult.However,Software-defined networking brings new ideas for providing virtual network services on heterogeneous networks.The physical network collects device information through the controller and s it as physical network resources for maintenance and management.When a virtual network request comes,it is mapped to part of the physical network through a virtual network mapping algorithm to meet business needs.In such a heterogeneous environment,if network resources cannot be properly configured,the resources of heterogeneous will be wasted,even worst,the overall performance of the network may be reduced due to heterogeneous network switching.Therefore,the thesis focuses on the virtual network mapping algorithm in the heterogeneous SDN scenario,aiming to realize an efficient and reasonable network resource allocation and virtualization scheme.Firstly,based on the existing virtual network mapping problem model,the thesis proposes a heuristic online network mapping algorithm that considers the remaining bandwidth of the link.In a heterogeneous network,when inter-network cooperative communication is performed between adjacent links,intermediate forwarding nodes will introduce protocol conversion overhead.The thesis first defines the cost of heterogeneous network switching in view of the traditional mapping model's insufficient consideration of heterogeneous network switching,and then introduces it into the physical resource expenditure model,and minimizes the physical network expenditure as the algorithm optimization goal.The physical resource consumption of the mapping scheme evaluated by the algorithm is more in line with the actual heterogeneous network model.At the same time,in view of the problem that the traditional link mapping scheme does not consider the multiple links between adjacent nodes,the thesis proposes a path relaxation algorithm,so that the algorithm can select the best link among multiple links.The simulation results show that compared with the existing VNE-G-SP,VNE-R-ViNE-SP and VNE-G-LMAoKS algorithms,the algorithm proposed in the thesis can improve the success rate of virtual network requests,reduce the resources allocated by the physical network,and improve the utilization of network resources.Secondly,the thesis proposes a virtual network online mapping algorithm based on ant colony algorithm improved bandwidth and delay awareness for the virtual network establishment request with multiple QoS requirements with both delay and bandwidth constraints.Aiming at the problem that the existing ant colony mapping algorithm is not suitable for heterogeneous networks in the ant random search process,the thesis introduces the remaining bandwidth resources and the heterogeneous switching cost into the definition of the heuristic function and pheromone model,so that the ants are more inclined to look for homogeneous links with abundant remaining resources.At the same time,the concept of expand graph in the same-stage mapping is introduced,so that each heterogeneous physical link can store pheromone,which makes it easier to approach the optimal solution.The simulation results show that compared with the VNE-Greedy and VNE-AC algorithms,the algorithm proposed in the thesis can improve the success rate of virtual network requests,optimize the ant colony search scheme,reduce physical network resource expenditures,and improve network resource utilization.Finally,the thesis summarizes all the work,discusses the research points that still need to be improved and looks forward to the next research work.