提出非齐次左—右型级联隐马尔可夫模型,用于应用层网络协议报文建模,描述状态之间的转移规律和各状态的内部相位变化规律,刻画报文的字段跳转规律和字段内的马尔可夫性质,基于最大似然概率准则确定协议关键词的长度,推断协议关键词,自动重构协议的报文格式。实验结果表明,所提出方法能有效地识别出协议关键词和重构协议报文格式。
A left-to-right inhomogeneous cascaded hidden Markov modelwas proposed and applied to model application protocol messages. The proposed modeldescribed the transition probabilities between states and the evolution rule of phases inside the states,revealed the transition feature ofmessage fields and the left-to-right Markov characteristicsinside the fields. The protocol keywords were inferred by selecting lengths with maximum likelihood probability, and then the message format was recovered. The experimental results demonstrated that the proposed method perform well in protocol keyword extraction and message format recovery.