Using Incremental Mining Approach to Analyze Network Traffic Online Based on Fuzzy Rules

Most of Network Intrusion Detection Systems (NIDSs) and network management applications need to analyze user's behavior from network traffic as soon as possible. Fuzzy association rule is one of the most popular approaches to describe user's behavior from network traffic. In the paper, we propose a fast fuzzy association rules generating algorithm that can complete once mining from milliseconds to seconds, by incremental mining approach. That is, our algorithm makes it possible to on-line analyze living packets by fuzzy association rules. Extensively experiments and analyses are done to show the performance of the algorithm, especially for the memory consumption and time cost.