一种基于改进隐马尔可夫的多媒体业务分类算法

摘要

This paper proposes an improved Hidden Markov Model (HMM) based multimedia traffic classification method. This method preserves the classical HMM model structure, and improves the performance of multimedia traffic classification by changing the emitting probability value with the position information of packet size. Theoretical analysis indicates that the new model can reduce the computational complexity of the classical HMM model. Simulation results show that the proposed method can improve the classification performance compared with the existing HMM based classification method.%该文提出一种基于改进隐马尔可夫(Hidden Markov Model, HMM)的多媒体业务分类算法。改进后的算法保持典型 HMM 模型结构不变，通过区分包大小的位置信息，改变发射概率取值，提高了多媒体业务区分性能。理论分析表明，该文模型在计算量上低于高阶HMM；实验结果表明，改进的HMM多媒体业务分类算法的区分效果优于现有的HMM多媒体业务分类方法。