Detection of Malicious Cloud Bandwidth Consumption in Cloud Computing Using Machine Learning Techniques.

Area of Science:

• Cybersecurity
• Computer Networks
• Distributed Systems

Background:

• The Internet of Things (IoT) presents a large attack surface due to hyperconnectivity and heterogeneity.
• IoT devices deployed in diverse environments are vulnerable to novel cyberattacks.
• Existing security measures struggle to autonomously detect and respond to IoT threats in real-time.

Purpose of the Study:

• To develop an intelligent self-protection security system for IoT environments.
• To integrate autonomous attack detection, prediction, and response mechanisms using fog computing.
• To enhance IoT security by minimizing human intervention and accelerating threat management.

Main Methods:

• Implemented a fog computing architecture with intelligent self-protection mechanisms in distributed fog nodes.
• Utilized distributed Gaussian process regression at fog nodes for predicting novel attack patterns.
• Employed fuzzy logic for selecting appropriate responses to detected and predicted attacks.

Main Results:

• The proposed system accurately detects and predicts both known and novel cyberattacks in IoT networks.
• Distributed Gaussian process regression demonstrated faster and more precise attack prediction in uncertain IoT environments.
• The fog computing-based system achieved a 25% faster threat neutralization rate compared to cloud-based implementations.

Conclusions:

• Fog computing enables effective self-protection for IoT applications by integrating intelligent mechanisms into fog nodes.
• The developed system provides autonomous, rapid, and accurate security responses, crucial for the dynamic IoT landscape.
• This approach significantly improves IoT security posture, reduces bandwidth usage, and enhances overall system resilience.