The flow of the application-level analysis system for HTTP
traffic is as follows. First, packets with the same 5 tuple
information are collected to create flow to determine if it is
HTTP traffic [6]. In the case of HTTP traffic, User-Agent
information is extracted and when User-Agent is normally
extracted, it is sent to signature-based HTTP application
analysis system [7]. When User-Agent is not normally
extracted, it performs the comparison with the information on
flow that was 1 minute ago. For example, in the case in which
User-Agent information does not exist in the flow created at
12:31, it compares with the information on flow that was
created a minute ago (12:30) and performs grouping with the
same flow when 5 tuple information are consistent and
performs mapping through application analyzed a minute ago.
This is a process for solving problems that occur when flow is
created in 1 minute interval, as mentioned in Section 2. If the 5
tuple information of current flow is consistent with the 5 tuple
information of flow that was created a minute ago, the two
flows were created in the same application and accordingly
saved as the same User-Agent. Table 1 shows the 5 tuple
information of flow created at 12:30 on February 20, 2013 and
the 5 tuple information of flow created at 12:31 on February 20,
2013.