Clustering-Based Anomaly Intrusion Detection Technology

Today, the information technology goes into every sphere of the society. People's working and daily life are increasingly rely on computer and network. But the problem of network security is cannot overlooked and challenges on people with network providing convenience to people. The network security problems not only involve in personal information security but the national information security as well. The dramatically developing attack technologies and malicious access, the security bug by nature caused by leaks of coding and design and substantial amount of computer viruses make the current security technologies such as firewall, identity authentication technology, operating system security kernel technology too weak to protect the network. The intrusion detection system with active protection, dynamical monitoring and system protection together have a developed at a rapid pace in recent decades. Furthermore, the intrusion detection system merging supervised and unsupervised machine learning algorithms have became the hot spot in relative research fields.This paper proposes a algorithm which can improve the accuracy and race of the intrusion detection system by studying on relative papers and academic resources from home and abroad.First, by calculating information gain of all features for every specific attack type, we delete the redundant and reduplicate features which are not play substantial roles in discrimination process from KDDCUP99.Secondly, we use the improved algorithm of Particle Swarm Optimization to improve FCM. In the step of initialization, we set the value of K 5. By using the improved FCM, we can get five clustering centroids. For each data point in dataset, we randomly choose two points from the five clustering centroids and use them and the data point to construct ten triangles. The we calculate the areas of ten triangles and set these areas as the new feature.Lastly, we use SVM to train and the intrusion detection model based on the new feature vectors and get the final experimental result. Our system achieve the accuracy of 95.88% and false alarm rate of 2.99. Meantime, we also achieve a better detection performance for unknown attack.