A Malicious Code Detection Method Based on FF-MICNN in the Internet of Things.

Area of Science:

• Computer Science
• Cybersecurity
• Machine Learning

Background:

• The security of the Internet of Things (IoT) is increasingly threatened by malicious code.
• Effective detection of malware is crucial for safeguarding IoT devices and networks.

Purpose of the Study:

• To propose a novel malicious code detection algorithm for IoT security.
• To enhance the accuracy and efficiency of malware detection through feature fusion and deep learning.

Main Methods:

• Malicious code is converted into grayscale image features using image technology.
• Opcode sequence features are extracted using the n-gram technique.
• Global and local features are fused, then processed by a feature fusion-malware image convolutional neural network (FF-MICNN) for training and classification.

Main Results:

• The proposed FF-MICNN algorithm demonstrates improved detection speed and feature comprehensiveness compared to existing methods.
• The algorithm achieved a 0.2% higher accuracy rate than detection algorithms relying on single features.
• Experimental results validate the effectiveness of the feature fusion approach in enhancing malware detection.

Conclusions:

• The FF-MICNN algorithm offers a promising solution for robust malicious code detection in IoT environments.
• Combining feature fusion with deep learning significantly improves malware detection performance.
• This research contributes to advancing the security measures for the rapidly growing Internet of Things ecosystem.