Behaviour-Based Malware Detection for Resource-Constrained IoT Devices

Resource-constrained IoT devices present a unique challenge as they lack the necessary processing power and memory to run traditional detection methods. In this paper, we explore the challenges of malware detection in such devices and propose a solution using ensemble machine learning models. Our research focuses on identifying the key features and behaviours for malware detection in IoT devices, aiming to improve detection effectiveness. Using the IoT-23 dataset, our main goal is to create a more advanced machine learning model capable of accurately categorizing network traffic as either malicious or benign. The dataset consists of both benign and malicious data, with various feature columns undergoing transformations through feature engineering techniques for model compatibility. Machine learning algorithms are employed in this research to enhance the precision of identifying harmful traffic, enabling more accurate detection. To enhance the precision of categorizing network traffic, one can optimize classification accuracy using a randomized classifier and adjusting parameters accordingly. The model's effectiveness can be assessed using a confusion matrix to evaluate outcomes.