Application of SDN Network Traffic Prediction Based on Speech Recognition in Educational Information Optimization Platform

Abstract

This paper constructs a SDN network traffic prediction model based on speech recognition and applies it to the educational information optimization platform. By analyzing the influencing factors of SDN network equipment, communication links, and network traffic, this paper constructs the initial index set of SDN network traffic situation. In the data plane of SDN, the queue management algorithm is used to control the flow. On this basis, an IRS mechanism is proposed based on the advantages of SDN centralized control and the difference of transmission performance requirements between large and small streams. For the transmission of large traffic, IRS adopts greedy routing and multipath routing based on the remaining bandwidth to make the traffic evenly distributed in the network, and IRS adds the scheduling strategy based on IP addressing to avoid packet disorder. Simulation results show that the effectiveness of this algorithm can reach 95.67% at the highest, and the MSE convergence is 0.0021 at the lowest. At the same time, this method completes the quantitative evaluation of SDN network traffic situation, effectively solves the problem that SDN traffic situation labels cannot be determined, and opens a new vision of global state observation for SDN network management. This research can provide some technical support for the educational information optimization platform.