Identifying Tampered Radio-Frequency Transmissions in LoRa Networks Using Machine Learning

Area of Science:

• * Internet of Things (IoT) security
• * Wireless communication systems
• * Cybersecurity and data integrity

Background:

• * Long-range (LoRa) networks are crucial for IoT, offering low-power, long-range communication.
• * LoRa devices are susceptible to radio-frequency interference and signal manipulation, compromising data integrity and security.
• * Detecting tampered frequency signals in LoRa networks is a significant challenge.

Purpose of the Study:

• * To present an innovative method for detecting tampered radio frequency transmissions in LoRa networks.
• * To evaluate the efficacy of five anomaly detection algorithms for identifying signal manipulation.
• * To enhance the security and reliability of LoRa-based IoT systems.

Main Methods:

• * Utilized five anomaly detection algorithms: Local Outlier Factor, Isolation Forest, Variational Autoencoder, traditional Autoencoder, and Principal Component Analysis.
• * Employed image-based tampered frequency techniques by converting LoRa transmission signals into images.
• * Generated a dataset of over 26,000 images from real-world experiments with normal and manipulated signals.

Main Results:

• * Local Outlier Factor (LOF) demonstrated the highest detection accuracy at 97.78%.
• * Variational Autoencoder (VAE), traditional Autoencoder (AE), and Principal Component Analysis (PCA) achieved 97.27% accuracy.
• * Isolation Forest (IF) achieved 84.49% accuracy in detecting tampered signals.

Conclusions:

• * The proposed image-based anomaly detection methods are effective in identifying tampered radio frequency signals in LoRa networks.
• * Local Outlier Factor shows superior performance in detecting signal manipulation.
• * These findings offer a promising approach to bolster the security and reliability of LoRa-based IoT infrastructures.