Mining Approximate Frequent Closed Flows over Packet Streams

Due to the varying and dynamic characteristics of network traffic, the analysis of traffic flows is of paramount importance for network security, accounting and traffic engineering. The problem of extracting knowledge from the traffic flows is known as the heavy-hitter issue. In this context, the main challenge consists in mining the traffic flows with high accuracy and limited memory consumption. In the aim of improving the accuracy of heavy-hitters identification while having a reasonable memory usage, we introduce a novel algorithm called ACLStream. The latter mines the approximate closed frequent patterns over a stream of packets. Carried out experiments showed that our proposed algorithm presents better performances compared to those of the pioneer known algorithms for heavy-hitters extraction over real network traffic traces.