| In order to cope with the explosive growth of data traffic and diversified service demands in the future,5G network slicing technology has emerged.With the help of Software Defined Network(SDN)and Network Function Virtualization(NFV)technology,network slicing is a technique for flexible resource allocation which cuts and regroups limited physical resources to form logically independent virtual network resources for each slice,therefore enabling centralized management and providing better quality for tenants.However,due to the limited physical resource,how to effectively deploy Virtual Network Function(VNF)and migrate VNF in the slices to meet the Quality of Service(QoS)requirement has become a critical issue in the virtualization research.The thesis focuses on the VNF deployment problem in 5G network slicing.The main contents and innovations of this thesis are summarized as follows:1.To solve the problem that the existing methods cannot achieve the optimization of end-to-end delay and reliability of Service Function Chaining(SFC)deployment simultaneously,a model of SFC deployment with QoS requirements is proposed,and further a dynamic deployment algorithm for SFC with QoS requirements is designed.The method evaluates the nodes based on the approach of Google PageRank through the perception of invulnerability of nodes and link reliability firstly.Then based on the principles of load balancing and coordination with link mapping,the VNF is mapped onto the physical node with the highest comprehensive resource capacity in the VNF deployment phase,which realizes the integral optimization of delay and reliability.Finally,the shortest path of delay that meets the reliability requirements is seclected for link mapping.Simulation results show that the proposed algorithm can reduce the end-to-end delay and ensure the reliability of SFC,as well as improve the request acceptance rate and resource utilization.2.To solve the problem of real-time migration caused by lacking effective prediction in the existing methods,a VNF migration algorithm based on Deep Belief Network(DBN)prediction of resource requirements is proposed.The algorithm firstly builds a system cost evaluation model integrating bandwidth cost and migration cost,and then designs a deep belief network prediction algorithm based on online learning which adopts adaptive learning rate and introduces multi-task learning mode to predict future resource requirements.Finally based on the prediction result as well as the perception of network topology and resources,VNFs are migrated to the physical nodes that meet resource threshold constraints through greedy selection algorithm with the goal to optimize system cost,and then a migration mechanism based on tabu search is proposed to further optimize the migration strategy.Simulation results show that the prediction model can obtain good performance and adaptive learning rate accelerates convergence speed of the training network.Moreover,the combination with the migration algorithm effectively reduces system cost and the number of service level agreements violations during the migration,so that the performance of network services is improved. |
