Research On Virtual Network Mapping Algorithm For Internet Of Things

With the popularity of heterogeneous networks for example 5G/6G as well as Io T,the activity of network consumers has seen increased steadily.in the meantime,the weak point of traditional Internet system based on old agreement are increasingly exposed in many aspects such as function expansion,mobile services,network security etc.Network virtualization technology solves these problems effectively by building a fully virtualized network structure,which make the management of network more flexible for operators.This thesis introduces the theory of network virtualization and the types of mapping algorithm.The focus of this chapter includes some special virtual network scenarios,for instance,cross domain as well as wireless and so on.Virtual network mapping problem is a combinatorial optimization problem.In order to deal with the mapping process better,polynomial time is taken as the symbol of complexity measurement,and the algorithm complexity under each type is calculated in the way of conversion.Two online virtual network mapping algorithms are designed for the scenario that the mapping process based on a single physical network whose nodes do not support segmentation.Both algorithms aim at maximizing the Persistent profitability of the Provider.The first algorithm focuses on the optimization of virtual node mapping process,the relative merits of genetic and harmony search algorithm were considered;and the initial harmony memory was formed by using the optimized population from genetic algorithm.In order to avoid group assimilation and premature convergence,the initial value of mutation probability is larger and decreases in a nonlinear way as iteration goes on.Scheduling resources in a fire way among multiple underlying physical nodes to realize the balanced utilization of resources really do count.The second algorithm focuses on the virtual link mapping stage,a time-varying algorithm based on the idea of load balancing was designed for the scenario of physical network supporting path segmentation.In the virtual link mapping stage,the unit charge coefficient used in the time-varying multi commodity flow model can reflect not only the load balance of the physical link,but also the consumption of it.We designed the physical network load balancing symbol u which simplified the multivariable maximum value problem.Finally,the performance of the two algorithms is analyzed relay on ns-allinone-2.35 Internet of things simulation platform,GT-ITM tool and MATLAB r2020 a.Both algorithms achieve the desired results.