Research And Analysis On Joint Optimization Mechanism Of Performance And Economics In Inter-Domain Traffic Engineering

The fast development and the expanding scale of the Internet have brought to the industrial community a huge challenge to network management and Internet technology. However, the closure of the inter-domain information and the selfishness of the inter-domain policy have led to greatly reduce of network utilization and great waste of network resources. So we need a new optimization mechanism in inter-domain traffic engineering to optimize the performance of the network, to use the network resources more effectively and to provide users with quality of service guarantees.Existing inter-domain traffic engineering is an optimization mechanism from the view of performance. The nature of inter-domain traffic engineering is a selection problem of the egress router, which leads the domain traffic to the best egress router within the autonomous domain. In this paper, we build the model to describe the interaction among the different roles in the network, and use game theory to capture the collaboration and competition mechanism among them. So we can optimize the model from the view of both performance and economics to find the best egress router for the domain traffic. In terms of performance, we use the Nash product as the optimization index, which has the Nash bargaining solution, being Pareto-efficient and provably fair. In terms of economics, we use the revenue sharing contract to make the two sides participated in stakeholders and work together to minimize the economic costs. In order to minimize the information leaked to other ISP, we decompose the master problem to sub-optimization problems for ISPs to solve independently, and finally provide a good egress router selection strategy to make both ISPs win-win. We evaluate our approach in several experiments with simulated networks. The experiment results show that our approach converges quickly, and have the same solution with the centralized solution with full knowledge of the networks. When compared with existing inter-domain traffic engineering methods, our approach optimizes performance and reduces economic costs at the same time, which is meaningful. Our approach is proved to be an effective solution to solve the problem of inter-domain traffic engineering.The paper is organized as follows. At first, the paper describes the background of the inter-domain traffic engineering problem and its related work. Secondly, we build the dual-objective optimization on performance and economics model according to the research goals. Next, we combine the Lagrangian decomposition algorithm and the dual-objective optimization algorithm to solve the problem, and use the adaptive Lagrangian relaxation algorithm to improve the convergence. Finally, we compare our approach with existing inter-domain traffic engineering methods to evaluate our approach on its effectivity and expansibility, and describe the future study goal.