Feed-forward ANN with Random Forest Technique for Identifying Malicious Internet of Things Network Intrusion

The Information Technology industry advancements over the recent years have contributed to a significant growth and development of smart and intelligent devices, which we now refer to as Internet of things (IoT). The IoT market continues to grow every year as manufacturers continue to create smart and intelligent devices meeting the needs of different industries, home users, and research experiments. While we experience a significant IoT growth, the security flaws in IoT design are threatened by attackers who target loopholes in IoT by designing botnets for exploit. In this work, the researchers proposed an approach for identifying botnets targeted to IoTs. The researchers proposed a set of activities to be followed on the BoTNeTIoT-L01 dataset. The initial step involved scaling the dataset using the most widely used min-max normalisation technique. An important step to avoid test data leakage. Following this, they applied multiple data analysis and dimensionality reduction techniques to achieve superior results. Experiments performed by the researchers shows the technique outperforms other IoT network intrusion identification approaches based on evaluation metrics used such as accuracy and computational efficiency.