An Investigation of Intrusion Detection in UDP Data Streams

Securing the local network is a crucial task for any system administrator, as the activities in a network vary widely from simple data searching to online commercial transactions in many organizations. Intrusion detection systems (IDS) are extremely useful in this task. Detecting intrusions and identifying various methods or types of intrusions play an important role for predicting an intrusion and securing the network in the long run. Data mining techniques are being increasingly used to study the data streams with good results in IDS. In this paper we propose to extract unique signatures from UDP data streams, and predict intrusion using data mining techniques. We have used the KDD cup 1999 dataset that contains a wide variety of intrusion attacks simulated in a military environment.