摘要
针对目前卫星网络的特点,提出一种基于机器学习的QoS路由机制。卫星网络节点路由的过程被看成是一个分类器分类的过程,其输入是一个包括路由的起始节点、目标节点、各QoS度量等参数的离散值向量,输出为1条最优路径的标志符。该机制利用本地流量的历史数据在实时网络拓扑下动态构造的训练集来训练分类器。为动态构造训练集,调用粒度受限算法在多项式时间内解决多约束QoS路由问题,而多约束QoS问题是一个NP完全问题。训练完成后,当收到实时QoS路由请求时,卫星节点根据分类器分类的结果进行路由。仿真结果表明:本文所提出方法在实际网络吞吐率高的背景下性能更加优越。
A QoS routing scheme based on machine learning was proposed to accommodate the features of satellite network.In this scheme,routing was deemed as the process of classification in a classifier.The input was a vector indicating the parameters of a QoS routing request,and the classifier output was a identifier of a optimal path.Local historical data reports were utilized to dynamically establish training sample sets applying current network topology,which were then used to train the classifiers.By invoking limited granularity heuristic algorithms,a multi-constrained QoS routing problem,which proved to be NP-complete problem,could be resolved in polynomial-time.After completing training,a test sample representing a real-time QoS routing request will be classified to a given path.The simulation results demonstrate that the scheme performs better than conventional source-routing schemes under the circumstances of high-throughput network in reality.
出处
《中南大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2013年第S2期263-267,共5页
Journal of Central South University:Science and Technology
基金
国家自然科学基金资助项目(60973145
61004021)
国家重点基础研究发展规划("973"计划)项目(2012CB821206)
关键词
卫星网络
多约束QOS
路由机制
粒度受限
机器学习
satellite network
multi-constrained QoS
routing scheme
limited granularity
machine learning