Deep learning-based improved transformer model on android malware detection and classification in internet of...

Area of Science:

• Cybersecurity
• Artificial Intelligence
• Automotive Engineering

Background:

• Autonomous vehicles (AVs) increasingly use the Android operating system, introducing significant security vulnerabilities.
• Android malware poses a substantial threat to the safety and functionality of AVs.
• Existing machine learning methods struggle to detect novel and complex Android malware variants effectively.

Purpose of the Study:

• To propose a novel Deep Learning-Based Improved Transformer Model on Android Malware Detection (DLBITM-AMD) for Internet vehicles (IoVs).
• To enhance the accuracy and effectiveness of Android malware detection in the context of autonomous driving.

Main Methods:

• The DLBITM-AMD technique employs Z-score normalization for data standardization.
• It utilizes Binary Grey Wolf Optimization (BGWO) for optimal feature subset selection.
• An improved transformer integrated with an RNN model and softmax, optimized by the Snake Optimizer Algorithm (SOA), is used for classification.

Main Results:

• The DLBITM-AMD method demonstrated superior performance in detecting Android malware.
• Extensive experiments on a benchmark dataset validated the technique's effectiveness.
• The proposed model achieved a high accuracy rate of 99.26%.

Conclusions:

• The DLBITM-AMD technique offers a robust and accurate solution for identifying Android malware in connected vehicles.
• This deep learning approach significantly improves upon existing malware recognition models.
• Implementing DLBITM-AMD can bolster the security and safety of autonomous driving systems.