Incentive-Driven Bidding Strategy for Brokers to Compete for Service Provisioning Tasks in Multi-Domain SD-EONs

It is known that the multi-broker-based management plane can potentially provide a realistic solution to facilitate incentive-driven cross-domain network orchestration in multi-domain software-defined elastic optical networks (SD-EONs). Such network orchestration assures the autonomy of each domain and supports economical service provisioning across multiple domains as well. In this study, we consider the economic principle in multi-broker-based multi-domain SD-EONs and study how to realize incentive-driven service provisioning with broker competitions. We first present the theoretical model of the network operations to describe the noncooperative game in which the brokers compete for interdomain provisioning tasks with only incomplete information on their competitors. Then, we analyze the Nash equilibrium in a simplified version of the game, and show that to maximize the brokers' profits in long-term repeated games, an effective bidding strategy is needed for the brokers to predict their competitors' behaviors and price their services in the optimal way. The bidding strategy is designed by leveraging the kernel density estimation scheme. Finally, to demonstrate the effectiveness of the proposed bidding strategy, we implement it in an OpenFlow-based multi-domain SD-EON control plane testbed. The experimental results verify that our system performs well and the brokers can obtain higher profits with the proposed bidding strategy in repeated games.