Anomaly-based network intrusion detection with ensemble classifiers and meta-heuristic scale (ECMHS) in traffic flow streams

The exponential growth in the internet services lead to enormous growth in the network traffic. As the services are increasing the numbers of network attacks are also gradually increasing. From the contemporary literature it is proved that machine learning techniques have gained importance in addressing security issues in networks and these techniques rely on features and its values to extract the knowledge. It is evidenced that phenomenal growth in the volume of transactions leads to deviation in feature values. Hence, it is necessary to consider the associability among the transactions and its feature values. In this paper, a Meta-heuristic association scale is proposed to derive a threshold value for the transaction and further, an ensemble classifier is used to analyse the transaction as normal or attack. Ensemble classifier used in the proposed system is based on drift detection which has the ability to analyze the requests at stream level. The proposed model derives the features from the stream level and uses drift detection to analyze the stream characteristics. The experimental study is carried out on the benchmark data to analyze the statistical parameters accuracy, false alarm rate, positive predictive values. Moreover, the ECMHS is compared with the other benchmark models depicted in contemporary literature.