Machine learning threat prediction and classification framework for IoT-enabled 5G networks

The proliferation of smart devices for a wide range of applications in fifth-generation (5G) networks has raised serious security concerns about network threats, especially at the perception, network, and application layers. The existing methods for Internet of Things (IoT) network security have made strides, but significant room for improvement remains in building defence mechanisms against attacks at the application and protocol levels. The problems of overhead implementation, single attack detection, and poor classification and detection approaches are pertinent. This paper aims to develop machine learning models for improved IoT network security through timely and accurate threat prediction. The objective is to use AdaBoost, decision trees, and artificial neural networks (ANN) to detect normal and threat patterns in IoT networks. We employed the wireless network intrusion detection (ID) dataset sourced from the NSL-KDD repository, which contains a total of 148,517 data points split into 125,973 training datasets and 22,544 testing datasets. Findings indicate that the decision tree (DT) outperformed AdaBoost and ANN with an accuracy of 99%, 84%, and 68%, respectively. However, deploying the framework in real-time network environments using fog computing platforms is recommended to determine its efficiency in real-world scenarios.