Simulating ML-Based Intrusion Detection System for Unmanned Aerial Vehicles (UAVs) using COOJA Simulator

We are living in an era of IoT devices and the rapid increase in the use of drone applications is evidence of that. UAVs or drones are being used in a variety of industries, ranging from military purposes to delivery purposes, they can be seen everywhere. UAVs come under the umbrella of Unmanned Aerial Systems (UAS). With the increased usage of drones, there is an increased number of cyber-attacks on drones as well. Previously, an IDS solution was developed using Random Forest Classifier algorithm and with help of the CIC-IDS2018 dataset for the identification of these emerging threats. This time, the target is to simulate a UAV environment using COOJA Simulator and evaluate the proposed IDS’ performance. IDS controller architecture is also proposed, which contains a parser, a selector, a router, machine learning algorithms, and a performance analyzer. The functionality of this controller is implemented inside the UAV motes. An SDN controller is used to manage the traffic between the UAV motes and generate malicious and benign traffic for IDS detection. The performance analysis module determines the performance of each algorithm. The best accuracy range was provided by Random Forest Classifier between 95%-96%.