ABSTRACT

with the increased use of internet for a wide range of activity from simple data search to online commercial transactions, securing the network is extremely important for any organization. Intrusion detection becomes extremely important to secure the network. Conventional techniques for intrusion detection have been successfully deployed, but predictive action can help in protecting the system in the long run. Data mining techniques are being increasingly used to study the data streams and good results have been achieved over time. In this paper we propose to extract unique signatures from UDP data stream, apply existing mining techniques and compare results. We have used the KDD cup 1999 dataset which contains a wide variety of intrusion attacks simulated in a military environment.