A study of anamoly detection in networking using soft computing techniques

Abstract

Now-a-days network security is an important field in protecting the communication networks from the cyber crime, cyber threats, unauthorized access, etc. The anomaly detection is one of the important techniques for identifying the anomalous patterns that do not establish the normal behaviour pattern. The main objective of this research is the application of soft computing techniques for solving the problem of anomaly detection in networking. Anomaly Detection (AD), in which the analysis looks for abnormal patterns of activity, has been, and continues to be the subject of a great deal of research. In this thesis the following three different soft computing techniques are used to solve anomaly detection in networking with improved the detection rate and false alarm rate: and; Artificial Neural Network (ANN) and; Artificial Immune System (AIS) and ; Hybrid AIS based Genetic Algorithm (GA) First, two types of ANN are used for anomaly detection in networking. One is Multi-layered feed forward (MLFF) neural network approach and the other is Radial Basis Function (RBF) neural network.In this research in addition to ANN, dimensionality reduction is also applied. There are two different approaches to achieve dimensionality reduction: Feature extraction and feature selection. In this research work, a Principal Component Analysis (PCA) based feature extraction and Mutual Information (MI) based feature selection are used for dimensionality reduction with MLFF and RBF neural networks.