Research On Virtual Network Embedding Algorithm Based On Particle Swarm Optimization

The rapid development of network technology makes the "rigid" problem in the structure of the Internet become increasingly prominent,and becomes the shackles of the Internet development.Network virtualization is a kind of emerging Internet technology,which does not have to change the original underlying physical network architecture;it makes the coexistence of multiple heterogeneous virtual networks possible,and gradually becomes an important factor in solving the future Internet architecture.As the key technology of network virtualization,the research of virtual network embedding has become the hotspot in the industry research in recent years.How to map the virtual network requests to the underlying physical network,and allocating resources to virtual network reasonably is the key of virtual network embedding problem.Virtual network embedding was proved to be NP-hard problem,so most researchers proposed efficient algorithms by limiting the solving space of the problem.In recent years,particle swarm optimization algorithm,genetic algorithms and other intelligent optimization algorithm in solving combinatorial optimization problems and the results are remarkable,which provides a new idea for exploring using the intelligent optimization algorithm to solve resource allocation problems in virtual network embedding.The main research contents of this paper are as follows:(1)A particle initialization allocation strategy is proposed.By studying the virtual network mapping algorithm based on particle swarm optimization,points out the shortage of randomly generated initial particles,and put forward a kind of particle initialization allocation strategy,thus reducing process of debris particles distribution,improve the efficiency of node mapping.The simulation results show that this method can effectively increase the acceptance rate of virtual network request and improve the long-term average return overhead of virtual network request.(2)SPS-VNE-PSO algorithm is proposed.The virtual network embedding algorithm can largely improve the underlying physical network resource utilization,when the underlying physical network supporting path splitting.Therefore,this paper proposes a virtual network embedding algorithm for the underlying physical network path splitting supported,referred to as the SPS-VNE-PSO algorithm.The experiments show that the SPS-VNE-PSO algorithm can improve the network long-term operation benefit and the virtual network request mapping acceptance rate,and reduce the time of the virtual network embedding.(3)HPTS-VNE-PSO algorithm is proposed.Tabu search algorithm tries to avoid duplication of the search by using the tabu list,expanding the scope of the search from horizontal level.And simulated annealing algorithm in the longitudinal level to prevent falling into local convergence,both with particle swarm optimization(PSO)is combined with the optimization algorithm can effectively avoid the premature convergence in virtual network embedding.In this paper,a virtual network embedding algorithm based on hybrid particle swarm optimization is proposed,namely HPTS-VNE-PSO algorithm.The effectiveness of the proposed algorithm is demonstrated by the experimental comparison of the cloud computing simulation platform CloudSim.