A Comprehensive Survey on Deep Learning-Based LoRa Radio Frequency Fingerprinting Identification

Area of Science:

• Wireless Communication
• Cybersecurity
• Machine Learning

Background:

• LoRa is crucial for low-power, long-range IoT communication.
• Security and device identification are major concerns for LoRa deployments.
• Existing cryptographic methods face resource and physical attack limitations.

Purpose of the Study:

• To provide a comprehensive survey of deep learning-based Radio Frequency Fingerprinting Identification (RFFI) for LoRa devices.
• To analyze RF fingerprinting techniques and hardware imperfections for emitter identification.
• To review deep learning algorithms, signal representations, and datasets for LoRa device classification.

Main Methods:

• Surveying state-of-the-art deep learning RFFI techniques for LoRa.
• Analyzing hardware imperfections for unique device signatures.
• Examining various deep learning models and LoRa signal representations.
• Reviewing existing LoRa RF signal datasets.

Main Results:

• Deep learning RFFI shows promise for secure LoRa device identification.
• Hardware imperfections provide exploitable intrinsic features for identification.
• Diverse deep learning algorithms and signal representations are applied.
• A comprehensive review of datasets, including hardware and signal details, is presented.

Conclusions:

• Deep learning-based RFFI is a viable alternative to cryptography for LoRa device identification.
• Challenges and future research opportunities in this domain are identified.