Research On End-to-End QoS Routing Optimization Based On Traffic Engineering

In recent years,with the rapid progress of network technology,the scale of users and services in the network are growing by leaps and bounds,especially more and more intelligent physical devices are being connected to the Internet of Things(IoTs).The increasing scale of network and the huge amount of data in the network bring great challenges to the quality of service(QoS)assurance of the network.Network providers are building data center networks to support and manage massive IoT devices and traffic.In the context of the rapidly evolving network landscape,traditional hop-based routing algorithms are proving to be insufficient in guaranteeing end-to-end QoS.To address this challenge,traffic engineering has emerged as a highly effective means of managing network traffic in an end-to-end manner,optimizing network performance and resource utilization,and ultimately delivering reliable end-to-end QoS.This thesis proposes a routing optimization framework based on traffic engineering principles,tailored to the specific requirements of IoT and data center networks,with the aim of ensuring end-to-end QoS guarantees.In this thesis,we first implement an end-to-end delay minimization routing algorithm based on traffic scheduling in traffic engineering.Endto-end delay is modeled for data networks and subsequently,an iterative end-to-end QoS routing algorithm is designed based on Dijkstra’s algorithm and flow-based QoS algorithm.Then,this thesis designs and builds a simulation platform based on discrete-event simulation,and compares the proposed algorithm with the shortest path algorithm through simulation to verify its advantages in QoS metrics such as average end-toend delay,network throughput and packet loss rate.This thesis then incorporates Software Defined Network(SDN)technology into the IoT to improve network utilization.SDN enables the network to make routing decisions based on network information by separating the control plane from the forwarding plane.This thesis considers SDN-IoT networks,where IoT devices are accessed in using Zigbee protocol,and proposes a machine learning-assisted minimum endto-end delay(MaMED)routing strategy for IoT monitoring services;and the scheme does not require a priori knowledge of the input traffic.Simulations are performed in different SDN-IoT scenarios,and the results show that the proposed routing strategy outperforms the Shortest Path First(SPF)algorithm in terms of end-to-end delay performance and is more stable when the traffic increases.Finally,this thesis proposes a multipath routing scheme under end-toend delay constraints.First,the data center network is modeled as a queuing network,and the multicommodity flow problem with end-to-end delay constraints for traffic groups is formulated as a nonlinear constrained optimization problem.The object of this problem is minimizing the average end-to-end delay of the network under the constraints of link capacity and end-to-end delay per traffic flow.Subsequently,an algorithm based on the Flow-Deviation(FD)method is proposed to solve this optimization problem and analyzed.Finally,this algorithm is used in simulations to route the data center network and is compared and analyzed with multipath routing strategies such as Equal Cost Multipathing(ECMP).