Exploring Machine Learning Applications for IoT-Based Honeypots: A Review

As the Internet of Things continues to expand in connectivity and scale, securing Internet of Things devices is an increasingly paramount issue. Many Internet of Things devices are situated in environments absent of any safeguarding protocols. Insecure systems possess vulnerabilities that provide unauthorized access to attackers outside of the network, putting sensitive data at risk of exploitation. In recent years, attacks targeting Internet of Things devices increased in complexity and frequency. In 2022 alone, there were over 112 million global Internet of Things cyberattacks, an 87% increase from the previous year, emphasizing the necessity for improved security measures. For this to occur, an understanding of the mechanisms behind targeted attacks is needed. A way in which cybersecurity professionals gather information on adversarial behavior is through the usage of the honeypot tool. A honeypot is a decoy system that is designed to bait attackers into exploiting the system. Honeypot will log and analyze subsequent attack methodologies, providing valuable information on how institutions can adjust strategy to better secure systems. Recent studies have proposed novel approaches which leverage machine learning techniques to enhance Internet of Things-based honeypot performance. This study examines these techniques and provides a comprehensive analysis on current trends, limitations, and future research directions.