Detection of data integrity attacks by constructing an effective intrusion detection system

Network users are heavily targeted by data integrity attacks that affect the development of new security techniques. The main challenge in network security is to identify this kind of attack for the improvement of growing mechanisms. In this paper, a data integrity based effective intrusion detection system (DI-EIDS) is constructed to prevent the network with a high detection rate and low false alarm rates. It is classified into two phases; data sampling and selection of features. In the data sampling process, attacks are detected and inference based on the sample signatures. In this process, the Deviation forest (d-forest) is used to remove barriers; Grey Wolf Optimization (GWO) is used for sampling ratio optimization and Black forest (BF) classifier to obtain the best training data. To select the best features, GWO and BF are repeatedly used. Finally, DI-EIDS based on the black forest is constructed using the best training data set obtained by data sampling and feature selection. Rare Integrity attacks are detected in this technique when compared with other algorithms. Experimental results are analyzed using different datasets with a 22% sampling rate. The performance results show a higher rate of detection with low false-positive rates.