Profit-Driven Multi-Broker Based Gaming In Multi-Domain SD-EONs

In recent years,with the rapid development of the Internet,the bandwidth require-ments of the users have been increased in an explosive way,which demands for the larger traffic capability,higher flexible and dynamic networking.Optical networks have been widely focused due to their advantages on high-speed data transmission.Elastic optical networks(EONs)adopt flexi-grid optical orthogonal frequency division multi-plexing(O-OFDM)to realize finer granularity(12.5GHz or less)spectrum allocation.By leveraging the idea of software-defined networking(SDN),software-defined elastic optical networks(SD-EONs)can be built,which can not only inherit the advantages of EONs but also facilitate a powerful network control and management(NC&M)mech-anism.Note that,the broker is introduced as a higher orchestrator than the controllers to set up cross-domain lightpaths in multi-domain SD-EONs.Then,the multi-broker based multi-domain SD-EONs are proposed to improve the survivability and reliability of cross-domain operations.Brokers calculate inter-domain provisioning schemes and decide bidding prices for the inter-domain lightpath requests submitted from controllers to gain profits,which makes the competitive relationships among brokers.Game theory is leveraged to describe the market-driven behaviors of these brokers.We concentrate on the non-cooperative gaming among brokers.Controllers sub-mit cross-domain lightpaths to brokers,while brokers calculate provisioning schemes as well as the prices,however,only the broker bidding the lowest price can be chosen as the winner to provision the lightpath in each game.Thus,it is very important for these market-driven brokers to decide the optimal prices to realize profit maximization.We utilize game theory to analyze the Nash equilibrium in a simple gaming scenario,and moreover,we propose a biding strategy for brokers in a more practical scenario.Specif-ically,each broker predicts the future bidding behaviours of its competitors according to their historical bidding records by leveraging Gaussian kernel density estimation(KDE)scheme to obtain the optimal price.Simulations are designed and conducted to verify that the competition among brokers facilitates more efficient resource utilization and higher profits for brokers.To further protect the interests of the brokers.we study the cooperative gaming among brokers based on their non-cooperative bidding.Inter-domain lightpath requests are allocated to brokers according to their expected profits by leveraging Nash bargain-ing,and each broker can only calculate provisioning schemes and decide bidding prices for the requests in its market share.Firstly,market partition engine is introduced to pro-vide the brokers with a cooperative placement.Then,model the inter-domain lightpath request allocation into a Nash bargaining problem and propose a heuristic algorithm to realize time-efficient collision-free request allocation.At last,simulations are con-ducted to verify the good performance of the proposed algorithm.