Obfuscated Memory Malware Detection in Resource-Constrained IoT Devices for Smart City Applications.

Area of Science:

• Cybersecurity
• Machine Learning
• Embedded Systems

Background:

• Obfuscated Memory Malware (OMM) poses a significant threat to interconnected systems like smart cities.
• Existing OMM detection methods are often binary-focused, limited in scope, and unsuitable for resource-constrained IoT devices.

Purpose of the Study:

• To propose a multiclass, lightweight malware detection method for embedded devices.
• To address the limitations of existing OMM detection techniques in terms of scope and resource requirements.

Main Methods:

• A hybrid model combining Convolutional Neural Networks (CNNs) for feature learning and Bidirectional Long Short-Term Memory (BiLSTM) for temporal modeling.
• Development of a compact and fast processing architecture suitable for IoT deployment.

Main Results:

• The proposed method demonstrates superior performance in detecting OMM and identifying specific attack types compared to existing machine learning models.
• Experimental validation using the CIC-Malmem-2022 OMM dataset confirms the method's effectiveness.

Conclusions:

• The developed hybrid model offers a robust and compact solution for defending against obfuscated malware on IoT devices.
• This approach enhances the security of smart city applications by enabling efficient malware detection in embedded systems.