Deep Learning for IoT Security - Frontiers in Big Data Journal

DLIS 2022

Artificial Intelligence

With the availability of high-speed internet and smart sensors, IoT applications like smart homes, smart cities, connected healthcare, smart vehicular network, smart retail and supply chain, etc., are emerging with rapid speed. It is estimated that there were more than 26.66 billion IoT devices active in 2020, and this is expected to grow to 75 billion by 2025. Every second, a massive amount of data is being generated, shared, and processed. IoT applications are changing how people learn and work, while heterogeneity, ubiquity, and a massive scale of IoT are exposing us to increasingly serious security threats at the same time. The security vulnerabilities in IoT are made evident by several high-profile hacks in recent years, including the late 2016 Mirai malware DDoS attacks and the 2017 casino fish tank IoT thermometer hacking.
With the popularity of big data, the use of deep learning has been growing in a wide range of cybersecurity applications like intrusion and malware detection, user authentication (biometrics), user privacy, etc. Deep learning can be used to process and learn from the underlying IoT data to improve the threat assessment and attack identification as well as recognize breaches within the IoT ecosystem. Deep learning can also be applied to identify advanced threats such as organization profiling, infrastructure vulnerabilities, and potential interdependent vulnerabilities and exploits. Deep learning can significantly change the cybersecurity landscape. For example, traditional signature-based techniques for malware detection cannot keep up with the pace of new attacks and variants. New attacks and sophisticated malware have been able to bypass network and end-point detection to deliver cyber-attacks at alarming rates. The enormous scale of the latest ransomware attacks continues to remind us how difficult it is to protect networks from being infiltrated. Deep learning can be leveraged to learn the new defense mechanisms using all available data and address the growing cybersecurity problem.
This Research Topic focuses on recent advances in research and development in securing the IoT landscape using deep learning. The objective of this collection is to bring together researchers from both deep learning and cybersecurity domains to provide a venue to share ideas and foster knowledge on IoT security challenges and solutions.
Papers can be from any of the following areas, including but not limited to:
● Vehicular ad-hoc network, smart home, healthcare, and smart meter security
● Real-time/anti-adversarial/efficient deep learning security protocols/solutions
● Intrusion detection and prevention
● Malware detection
● Data security and privacy
● Anomaly detection in authentication, authorization, and data requests
● Adversarial machine learning and the robustness ofAI models against malicious actions
● Interpretability and explainability of deep learning models
● Privacy-preserving deep learning algorithms
● Trustworthy deep learning
● Deep graph learning
● Fog-based IoT security
● Sensor network security
● Cloud-based IoT security