MLAPS: A Machine Learning based Second Line of Defense for Attack Prevention in IoT Network

Intrusion Prevention Systems (IPSs) are processes that are handled as add-ons to the Intrusion Detection Systems (IDSs) in order to effectively resist and prevent invasions in IoT devices. In this paper, a comparative study on various Machine learning (ML) based Intrusion Prevention System (IPS) has been proposed on UNSW-NB15 dataset by applying various ML methods for classification. For Intrusion Prevention System, two level classifiers for dropping of packets are implemented. Level 1 classifier comprises of Decision Tree (DT) whereas Level 2 classifier uses Random Forest (RF). The proposed IPS automatically adapts to system changes and further reduces training time and increases the accuracy of the system and acts as a lightweight Second Line of Defense System for an IoT Network. In this system, threshold value is calculated by finding the average time of execution of all of the data in the training set.