Botnet Attacks Detection Using Embedded Feature Selection Methods for Secure IOMT Environment

Abstract

AbstractSecurity remains a primary responsibility, despite the expanding number of IoT devices. The IoMT (Internet of Medical Things) ecosystem is subdivsion of IoT which is made up of medical computing devices, software applications, and healthcare systems. The most critical criteria in an IoMT (Internet of Medical Things) environment are privacy and security. A security breach in a healthcare network might directly result in the death of patients, the Internet of Things (IoMT) necessitates enhanced protection. A botnet is a collection of internet-connected devices used to carry out security breaches, leak sensitive information, and give botnet attackers control over IoMT devices. Malware distributed via botnets often includes network communication features that allow attackers to connect with other threat actors via the botnet’s vast network of infected devices. Using the UNSW-NB15 Dataset, we proposed a ESLRFBM (Efficient and Secured LASSO and RFBM) model for predicting botnet attacks in an IoMT scenario. In the UNSW-NB-15 dataset, we employed the embedded feature selection methods LASSO and RIDGE to detect botnet attacks. To detect security concerns, the IOMT has evolved and used machine learning technologies. We created and evaluated performance metrics such as Hamming loss and AUCROC scores using both classic and hybrid classifiers. By comparing and analysing the results using LASSO and RFBM properties, we can determine that our described model performs better.KeywordsIoMT (Internet of Medical Things)Botnet AttacksDoSRIDGELASSOSecurityHealth care systems