AI-Driven Cybersecurity in IoT: Adaptive Malware Detection and Lightweight Encryption via TRIM-SEC Framework.

Area of Science:

• Cybersecurity
• Internet of Things (IoT)
• Machine Learning

Background:

• IoT devices face increasing sophisticated malware and zero-day attacks.
• Traditional security methods are inadequate for resource-constrained IoT systems.
• Real-time threat detection and secure communication are critical for IoT.

Purpose of the Study:

• To propose TRIM-SEC, a lightweight and scalable framework for unified threat detection and secure data transmission in IoT.
• To enhance malware detection accuracy and reduce false alarms in IoT environments.
• To provide efficient and secure communication for next-generation IoT ecosystems.

Main Methods:

• Autoencoder-Based Feature Denoising (AEFD) for data quality enhancement.
• Principal Component Analysis (PCA) for dimensionality reduction.
• Transformer-Augmented Neural Network (TANN) for malware classification using self-attention.
• Lightweight Elliptic Curve Cryptography (LECC) with Particle Swarm Optimization (PSO) for key generation.

Main Results:

• TRIM-SEC demonstrated higher detection accuracy compared to baseline models.
• The framework achieved a reduction in false alarms.
• Lower encryption latency was observed, suitable for real-time IoT applications.
• TRIM-SEC balances detection performance, cryptographic strength, and computational efficiency.

Conclusions:

• TRIM-SEC is a promising solution for securing IoT environments against advanced cyber threats.
• The framework's lightweight and scalable design is well-suited for resource-constrained IoT systems.
• TRIM-SEC effectively unifies intelligent threat detection with secure data transmission for smart ecosystems.