Efficient Ransomware Detection in Resource-Constrained Environments Using Optimized Multi-Layer Perceptron Networks
Tien-Chu Lin, Rongliang Lin
Abstract
Ransomware attacks represent a significant cybersecurity threat, particularly in resource-constrained environments such as the Internet of Things (IoT) and edge computing systems. Traditional detection methods face considerable challenges in these environments due to limited computational resources, which can hinder effective detection and response. To address these challenges, an optimized Multi-Layer Perceptron (MLP) model is proposed for real-time ransomware detection. The model is evaluated on two datasets: the RansomwareData dataset and the Ransomware_headers dataset. It achieves 97% accuracy and a 97% F1-score on the RansomwareData dataset. On the Ransomware_headers dataset, it reaches 96% accuracy and a 96% F1-score. Furthermore, the model demonstrates minimal resource consumption, using only 5.37% of CPU and 7.2 MB of memory, making it suitable for deployment in systems with limited resources. The proposed solution offers a reliable and efficient approach to ransomware detection. Future research will focus on enhancing the model's robustness against adversarial attacks, exploring ensemble learning techniques, and integrating lightweight, graph-based models.