Identification of Botnets and Nodes Attacking Smart Cities by Majority Voting Mechanism and Feature Selection

A network intrusion detection system can monitor network traffic, record it, and apply intrusion detection algorithms to detect network intrusions. An effective intrusion detection system must have the ability to learn and discover intrusion patterns, and for this reason, most intrusion detection systems are designed by machine learning and deep learning methods. In this article, an intrusion detection system in the fog layer of smart cities is presented with feature selection as a group and learning with majority voting to discover the important features of attacks and share them between fog nodes. Here, each fog node can detect the important characteristics of attacks and share it with other fog nodes. Each fog node can perform its own learning on the discovered important features to detect network intrusion. In the proposed method, each fog node recognizes the most important features of the network traffic by the feature selection mechanism as a group and by majority voting and uses these features to use the artificial neural network classifier. Simulation results show that the proposed method has accuracy, sensitivity, and precision of 98.89%, 98.68%, and 98.81%, respectively, in detecting network attacks. The proposed method is more accurate in detecting attacks than methods such as GTO, PSO, HHO, WOA, and JSO. Experiments showed that the proposed method has improved the accuracy index by 1.22%, 1.13%, and 0.77%, respectively, compared to the WOA, HHO, and JSO algorithms.