Research On Dynamic Resource Optimization Algorithm Of End-to-End Network Slicing

In recent years,Internet of Things(IoTs)has been developed rapidly,which bring new opportunities to traditional businesses,e.g.industry,agriculture,healthcare,security and so on.However,because of the diverse business and fragmented standards of Io Ts and vertical industries,it's impossible to use the same physical network to fulfill the varying requirements of different Io T business.To satisfy the demand of diverse business,Network Slicing(NS)based on Software Defined Networking(SDN)and Network Function Virtualization(NFV)has attracted enormous attentions,and becomes a key enabling technology of 5th-Generation Mobile Communication(5G)technologies.This paper mainly studies the dynamic optimization algorithm of end-to-end NS.We first develop a testbed based on dynamic resource allocation scheme of end-to-end NS,and then verifies the validity of the joint 3-dimentional resource optimization algorithm studied of computing,storage and communication.The main contents of this papers are summarized as follows:This paper researches on a dynamic resource allocation scheme of end-to-end NS,and builds a testbed on the basis of the scheme.We put forward detailed designs of the central controller,the local controller and the end-to-end NSs according to key technologies such as SDN and NFV,and the workflow of the proposed scheme.Then,on the basis of open source software and hardware,a testbed of dynamic resource allocation scheme on end-to-end NS is built.Finally,the scheme is tested and verified on the testbed.The test results show that the proposed scheme can dynamically allocate resources according to different business,and further achieve flexible resource management while guaranteeing users' quality of service(QoS).This paper studies a joint resource optimization algorithm of computing,storage and communication and verifies the performance of the algorithm on the testbed mentioned above.In specific,this paper models the joint resource allocation problem as optimizing the system throughput constrained on users' Qo S requirements.However,the problem is Nondeterministic Polynomial Hard(NP-hard),which means it's hard to derived a closed-form solution through traditional optimization methods.As a result,this paper puts forward a joint simulated annealing(SA)and particle swarm optimization(PSO)algorithm to solve the optimization problem.On one hand,PSO algorithm has the advantage of fast convergence;on the other hand,SA algorithm is introduced to avoid trapping in local optimum,which is the weakness of PSO.Finally,the experimental results show that the joint 3-dimentional resource optimization algorithm studied can realize fast iteration according to network requirements of different business and has good convergence.As a result,the new demands after the business change on the end-to-end NSs can be well satisfied.