KEAMANAN SIBER DALAM ERA INTERNET OF THINGS: TANTANGAN DAN SOLUSI TEKNOLOGI TERKINI

Abstract

The rapid expansion of the Internet of Things (IoT) has transformed industries but also heightened cybersecurity vulnerabilities. Cyber threats, including ransomware, data breaches, and distributed denial-of-service (DDoS) attacks, increasingly jeopardize critical infrastructure. Traditional security methods, such as encryption and firewalls, often fail to counter evolving AI-driven threats. This study introduces an AI-based security model that integrates deep learning and federated learning for real-time IoT threat detection and mitigation. The proposed system employs a hybrid CNN-LSTM architecture to analyze network traffic, while federated learning enhances detection accuracy and ensures data privacy. Experimental results demonstrate 92% detection accuracy, 4.2% false positive rate, and latency under 50 ms, outperforming conventional rule-based systems. Additionally, integrating AI with IoT protocols like MQTT and CoAP optimizes processing for low-power devices. The study highlights regulatory challenges, as 73% of industrial organizations lack AI-driven security policies. The proposed framework aligns with NIST SP 800-82 and GDPR, ensuring scalable and adaptive industrial cybersecurity solutions. These findings contribute to developing AI-driven security strategies, providing a foundation for enhancing IoT resilience against evolving cyber threats