Classification of Web Applications Using AiFlow Features

With the increasing of complexity and volume of network traffic, more advanced methods of traffic analysis are required to continue producing useful results. Conventional methods of solving this problem include lexicographical pattern analysis, where a signature is created manually and then compared with incoming traffic in the hopes of detecting a matching signature. The main issue of such methods is its inability to adapt to even minor changes in the signature of the target application architecture. In this paper we introduce a new flow format AiFlow, designed specifically to assist in the association of traffic with application based on a wider set of criteria. This flow format coupled with a sufficient artificial intelligence (AI), could be capable of identifying both the dynamic and static elements that define the behavior of a network-enabled application. Further, the system would be equipped to adapt to the inevitable variations in application behavior over time.