Empowering IoT with Large Language Models: A Survey of Applications, Challenges, and Future Directions

Abstract

The integration of Large Language Models (LLMs) with the Internet of Things (IoT) is reshaping intelligent systems, enabling advanced automation, cybersecurity, and natural user interaction in cyber-physical environments. This survey synthesizes recent advances in LLM-IoT integration, highlighting transformative applications, challenges, and emerging solutions. Key applications include intelligent automation, with AutoIoT achieving 94.1%–98.5% accuracy in zero-code orchestration; personalized healthcare, with LLM-HAS reducing false alarms by 8.147%; and cybersecurity via ChatIoT and BARTPredict, the latter with 98% intrusion detection accuracy. Challenges include computational constraints on edge devices, privacy risks in data handling, and hallucinations, with medical LLMs showing up to 15% error rates without fine-tuning. Solutions like federated learning, model compression (reducing memory usage by up to 50%), and hybrid frameworks such as LLMind improve efficiency and privacy. This survey offers a comparative analysis of these approaches, providing insights into their trade-offs. Future research should focus on multimodal LLMs, scalable edge deployments, and trustworthy systems powered by explainable AI. By addressing these issues, LLM-IoT ecosystems can evolve toward resilient, adaptive, and human-centric intelligent systems, enabling more generalizable AI in IoT-driven environments.

Ähnliche Arbeiten