Enhancing accuracy through ensemble based machine learning for intrusion detection and privacy preservation over the network of smart cities

The world is at peak of fifth-generation communication technology and adopting ideas like cloudification or virtualization, but, the most important element is still “security”, since more and more data is connected to the internet. Threat attacks are increasing in recent years, but classic network intrusion detection system has significant limitations that make it challenging to identify new attacks quickly. In this study, various supervised machine learning algorithms for anomaly-based detection methods are compared. The dataset utilized for anomaly-based detection techniques is KDDCup99 dataset, on which the different algorithms have been applied. The goal is to gain knowledge about data integrity and improve the predictive power of data. Given its ability to safeguard the integrity of data storage and maintain transparency in processes, this technology has promise for application in the field of intrusion detection. This study presents a technique for measuring various data parameters in network like accuracy, error rate, confusion matrix, etc. By accomplishing this, the amount of malicious data floating around in network can be reduced, making it a safe environment for data sharing. The accuracy of the proposed technique was found to be 99.82% accurate using the KDDCup99 dataset.