The invention discloses an unknown traffic clustering identification method based on an epsilon
machine, and the method comprises the steps: S1, carrying out the adaptive sampling of network traffic, and obtaining a traffic data sample; s2, extracting a feature sequence from the traffic data sample; s3, converting the feature sequence into a symbol sequence; s4, constructing an epsilon
machine of the symbol sequence by executing a causal state segmentation
reconstruction algorithm on the symbol sequence, and obtaining a causal state set corresponding to the epsilon
machine; s5, on the basis of the causal state set, calculating and outputting a
feature vector representing a flow mode; s6, calculating the distance between different feature vectors, and recording the distance as a
similarity distance; and S7, comparing the
similarity distance with a preset threshold value, and judging the type of a flow mode according to a comparison result to complete clustering identification of unknown flow. According to the invention, efficient and accurate identification of unknown traffic can be realized, and the invention also provides an
application firewall system and a computer readable storage medium, which also have the above beneficial effects.