A Machine Learning-Based Intelligent Framework With Two-Step Process For Leveraging Cybersecurity In Internet Of Things Use Cases

Abstract

With the growing use of Internet of Things (IoT) applications, it's crucial to prioritize security, as IoT devices are limited in resources and susceptible to various attacks. Currently, global security standards for IoT architectures are not widely enforced. Traditional security approaches are not feasible due to the limited computational power and energy of IoT devices. Artificial intelligence (AI) has emerged as a technology that innovates solutions in various domains. Machine learning (ML) is widely used to solve real-world problems in AI. It is a learning-based approach that incrementally gains knowledge to enhance cybersecurity. Existing literature on IoT application security mainly focuses on attack detection and classification. In a recent paper, we proposed an ML framework with a two-step process to improve IoT application security. In the first step, we utilized ML models for attack detection. In the second step, we used the best-performing model to classify attacks by leveraging labels in the process. Our proposed algorithm, Learning-based Cyberattack Detection and Classification (LbCDC), was tested using the UNSW-NB15 dataset. The experimental results showed that our system could detect and classify cyberattacks, achieving 89.55% accuracy for attack detection and 95.97% accuracy for attack classification. This ML framework can be integrated into organizations' existing security platforms.