Research On Resource Allocation Technology In Virtual Optiacal Network

With the rapid development of the Internet and the generalized used of cloud computing and big data technology nowadays,traditional Internet architecture is far more from satisfying the needs of the flourishing business at present,at which the virtual network virtualization technology emerges.Virtual network virtualization technology can operate virtual network requests from a variety of network architecture on the same substrate network,enabling the coexistence of the virtual network requests and the convenient deployment of the virtual network requests.This thesis mainly focuses on resource allocation technology in virtual optical network embedding situation and proposes two novel virtual optical network embedding algorithms aiming at reducing the maximum frequency slot on physical link and maximizing the revenue to cost ratio.The simulation results show that proposed algorithms achieves better performance in maximum frequency slot on physical link and the revenue to cost ratio than that of the comparison algorithms.First of all,this thesis proposes a virtual optical network embedding algorithm based on cooperation between node and link and randomized trial.Before choosing the mapping physical node in node mapping stage,the random trials are carried out.Physical mapping node is chosen according to the feedback value of the random trials.By taking the advantage of cooperation between node and link idea in random trails,our algorithm enables the distance between the mapping physical nodes are relatively close,and therefore,reducing spectrum resources in link mapping stage.Simulation results show that the virtual optical network embedding algorithm based on cooperation between node and link and randomized trial can achieve better performance in maximum frequency slot on physical link and the revenue to cost ratio than that of the comparison algorithms.Secondly,due to the great blindness of the previously proposed algorithm in the search process,it is not easy to find the optimal solution when the search space is large.Monte Carlo Tree Search continuously adjusts its search strategy according to the feedback results during the search process,making the searching direction continuously going in the direction of increasing the feedback value and approach the optimal solution.Therefore,drawing the advantages from Monte Carlo tree search and cooperation between node and link idea,this thesis proposes a virtual optical network embedding algorithm based on cooperation between node and link and Monte Carlo tree search.Combining the advantages of Monte Carlo tree search and the idea of cooperation between node and link,it can still achieve better performance when the search space is large.The simulation results show that the performance of virtual optical network embedding algorithm based on cooperation between node and link and Monte Carlo tree search is better than that of virtual optical network embedding algorithm based on cooperation between node and link and random trial.