基于BPNN端到端时延预测的多路传输调度  

Multipath scheduling based on BPNN for end to end transmission delay prediction

作　　者：苏旭东 衷璐洁[1] SU Xu-dong;ZHONG Lu-jie(Information Engineering College,Capital Normal University,Beijing 100048,China)

机构地区：[1]首都师范大学信息工程学院,北京100048

出　　处：《计算机工程与设计》2022年第1期66-72,共7页Computer Engineering and Design

基　　金：国家自然科学基金项目(61872253)。

摘　　要：针对移动异构网络环境网络参数动态变化、多路传输过程中数据包乱序引发的吞吐量下降等问题,提出一种基于BP神经网络(back propagation neural network,BPNN)端到端时延预测的多路传输调度方法,通过BP神经网络的构建、训练和学习,实现对端到端传输时延的更准确预测,以此为基础,对子流拥塞状况及网络状态进行综合评估后实施数据调度。仿真结果表明,该方法可有效减少数据包乱序的发生,实现负载均衡,提高网络吞吐量。Aiming at the problem of throughput degradation caused by out-of-order packets in the process of multipath transmission with the dynamic changes of network parameters in mobile heterogeneous network,a multi-path transmission scheduling method based on BP neural network(back propagation neural network,BPNN)end-to-end transmission delay prediction was proposed.Through the constructing,training and learning of BP neural network,more accurate prediction of end-to-end transmission delay was realized.On this basis,data scheduling was implemented after comprehensive evaluation of subflow congestion status and network status.Simulation results show that the proposed method can effectively reduce the occurrence of packet disorder,realize load balancing and improve network throughput.

关 键 词：多路传输 数据包乱序 数据调度 端到端传输时延 BP神经网络 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]