Research On WMN Routing Technology Based On SDN

Wireless Mesh Network(WMN)is a dynamic,multi-hop wireless self-organizing network.In WMN,the routing protocol is always one of its research priorities.So far,there have been various types of self-organizing routing protocols.However,these routing protocols have their own suitable application fields,and there is no routing protocol that can be applied to all present scenarios.Actually,in a traditional network,since the network topology is basically stable,a single routing protocol can meet the requirements.However,the network environment of WMN is changing rapidly,and a single routing protocol has become increasingly unable to meet the demand.So a solution that can adaptively determine the optimal routing protocol and deliver it to the underlying network according to the current network environment is needed.In order to solve these problems faced by WMN,some scholars have proposed to integrate the Software Definition Network(SDN)design ideas into the WMN network architecture,and proposed the concept of Software Defined Wireless Mesh Network(SD-WMN).SDN is a new network architecture that can separate the forwarding plane from the control plane to implement virtualization of the underlying network device and supports application development based on the global network.The addition of SDN design ideas has added greater scalability,dynamics,and manageability to WMN.It also provides a new solution for the WMN network to dynamically switch routing protocols based on the existing network environment.Based on the design concept of SDN and the project experience gained during the postgraduate study,this thesis proposes an adaptive routing protocol shearing scheme.The main contents and contributions are as follows:(1)This thesis verifies the theoretical feasibility of the proposed scheme by combining physical verification and simulation.Firstly,we realized the SD-WMN network architecture in the real environment through the physical verification method,and verified the feasibility of the SD-WMN network architecture.In addition,in order to meet the needs of rapid decision-making,this thesis proposes to apply the idea of machine learning to the decision of routing protocol,and realize the choice of fast routing protocol.This thesis collects thousands of network state data and corresponding performance parameters in the simulation environment,and uses the improved optimal routing protocol evaluation algorithm to label the original data.Finally,comparing the performance of multiple machine learning classification algorithms in optimal routing protocol decision,the random forest classification algorithm is selected as the best prediction model in the optimal routing protocol decision mechanism.(2)In order to further verify the decision-making results of machine learning,this thesis implemented the routing protocol change mechanism in the simulation software.The feasibility of the routing protocol change mechanism is proved by simulation experiments.And compare the delay,throughput,and packet loss rate before and after changing routing protocols to analyze the quality of machine learning decision-making.The experimental results show that the adaptive routing shearing scheme proposed in this thesis does choose a routing protocol with the best network performance and successfully implements routing protocol switching,which improves the overall network performance.