Study Of Internet Intra-domain Routing Optimization For Traffic Engineering

As with the rapid development of Internet,the increasing Internet applications and scales bring explosively growing Internet traffic.Such huge amount of traffic not only poses great challenges for improving network performance and quality of service,but also poses challenges for efficiently utilizing network resources.The goal of traffic engineering is to balance flows and make the most of network resources through routing optimization with limited network resources.However,due to the limitation of traditional routing protocols and routing algorithms,the problems on imbalance of network traffic? the degradation of quality of network service? the under-utilized of network resources need to be solved.Facing practical network scenarios and requirements,Internet intra-domain routing optimization for traffic engineering encounters great challenges in the determination of hybrid Software Defined Networking routing,the determination of coflow routing in wide area network,the determination of routing for different priorities traffic in wide area network.Focusing on the significant problem on Internet intra-domain routing optimization for traffic engineering,this paper studies Internet intra-domain routing algorithms for traffic engineering under the practical requirements of different network scenarios.The main content and contributions of this dissertation are shown as follows:1.Focusing on the intra-domain routing optimization problems and incremental de-ployment problem in a hybrid Software Defined Networking with a combination ofcentralized and distributed control,this paper proposes routing optimization algo-rithms to jointly optimize link weights of distributed routing protocol and splittingratio of centralized nodes,and proposes heuristic incremental deployment algo-rithms.For routing optimization problem,through building a mixed integer non-linear programming model based on multi-commodity flow problem,this paperproposes heuristic algorithms to adjust link weights of distributed routing protocoland exploits linear programming solver to obtain the splitting ratio of centralizednodes.For incremental deployment problem,this paper proposes heuristic incre-mental deployment algorithms to determine the deployment ratio and positions ofdeployed centralized nodes.Extensive experiments demonstrate the superiority ofthe proposed algorithms in reducing maximum link utilization and improving net-work performance.2.Focusing on the intra-domain routing optimization for task-level coflow in a widearea network,this paper proposes algorithms to jointly optimize task placementand coflow routing for a single coflow and multiple coflows.Through establishingthe mixed integer nonlinear programming model with semantic relation of flows,this paper proposes an approximation algorithm with performance guarantee to ob-tain the task placement and routing policy.Extensive experiments demonstrate thesuperiority of the proposed algorithms in reducing coflow completion time and im-proving quality of service.3.Focusing on the intra-domain routing optimization of different priorities traffic ina wide area network and congestion-free network update problem,this paper pro-poses a routing optimization algorithm to jointly optimize routing and the egressselection,and proposes a network update algorithm based on linear programming.For routing optimization,this paper builds a mixed integer nonlinear programmingmodel related to traffic of different priorities.This paper proposes a heuristic al-gorithm based on genetic algorithm to determine the egresses to Internet serviceprovider networks for the high priority traffic and the routing policies for trafficof different priorities.For the congestion-free network update problem,this paperproposes a network update algorithm based on linear programming to accomplishthe network update with minimal network state changes.Extensive experimentsdemonstrate the superiority of the proposed algorithms in improving network uti-lization and efficiently accomplishing a congestion-free network update.In summary,this paper models the problems of Internet intra-domain routing optimization for traffic engineering in different scenarios and proposes heuristic algorithms or approximation algorithms to solve the problems.Extensive simulation and testbed experiments demonstrate the correctness and efficiency of the proposed algorithms.