光互联网中的业务感知和波长调度机制研究

Research on traffic identification and wavelength scheduling scheme in optical internet

提出了使用广义到达时间间隔的贝叶斯网络业务分类感知算法和基于业务感知的光互联网波长调度机制。仿真结果表明,使用广义到达时间间隔的贝叶斯网络业务感知算法获得的平均感知准确率可高达98%,同时训练数据中业务类型所占比例与该类型业务的感知准确率有密切关系。基于业务感知的波长调度机制在获得更高波长资源利用率的同时,能够实现光网络资源与业务需求的更好适配。

A conception of universal inter-arrival time （UIAT）, and a traffic identification algorithm based on UIAT by using Bayesian network was presented. A novel wavelength scheduling scheme with traffic identification was proposed and numerically simulated. Simulation results show, the average traffic identification accuracy of the method is higher than 98%, and the identification accuracies vary with the number of instances of that traffic in the training data. It is also shown that the traffic identification based wavelength scheduling scheme can match the wavelength utilization with application requirements better with higher wavelength efficiency.