Research On Bottleneck Game Theory-based User-defined Routing System

Nowadays is the era of rapid development of Internet applications.While the network brings convenience to human beings,the continuously growing network traffic makes the existing network bandwidth resources more and more strained,leading to numerous problems such as network congestion.As an efficient network management tool,traffic engineering can provide scientific planning and flexible scheduling of network traffic,thus ensuring the improvement of network resource utilization.The traditional network architecture limits the development of traffic engineering,and the SDN architecture makes it possible to centralize the fine-grained control of traffic,making route planning more flexible and efficient.But the centralized computational requirements of the control plane can overload the hardware.In order to reduce the complexity of centralized traffic scheduling algorithms in SDN,and for the network to obtain user requirements more accurately and in real time,this paper then proposes a mechanism for user-network cooperation.This mechanism can be described as: empowering users to participate in network routing,and routing in the network is done in a distributed manner on the user side? while the network is mainly used to provide feedback to users on link status and implement their routing policies,thus solving the problem of excessive complexity of centralized algorithms.In order to realize this mechanism of user-network cooperation in scheduling traffic,this paper proposes a network architecture based on SDN technology that enables user-defined routing.After that,a prototype of user-defined routing system is implemented using Ryu controller and Mininet network simulation platform,and a functional simulation is completed.This paper verifies the feasibility of user-network cooperation mechanism and the performance improvement in load balancing.Existing traffic scheduling algorithms rarely take into account the performance impact of routing bottlenecks on the network.In fact,the performance of traffic routing depends on the performance of bottleneck links.At the same time,users compete for resources on the bottleneck links to form a game.Accordingly,this paper models the user’s traffic scheduling problem as a splittable bottleneck game and derives the social optimal case,thus designing a distributed routing algorithm based on best response.Later,a routing scheme based on link pricing strategy is proposed in the framework of cooperative routing between users and networks.The main idea of this routing scheme is to customize the rules for sharing the usage cost based on the link utilization rate,and to guide the users to allocate traffic to the less congested links through the link pricing policy,so as to finally achieve the win-win objective of user-perceived service quality improvement and network load balancing.Finally,the simulation and analysis work of the algorithm based on the network instances provided by SNDlib database is carried out in this paper,and the results show that the game-theoretic based algorithm has better performance in load balancing compared with Hedera and ECMP.