Considerations for Radio Frequency Fingerprinting across Multiple Frequency Channels.

Jose A Gutierrez Del Arroyo,Michael A Temple,Brett J Borghetti

doi:10.3390/s22062111

Jose A Gutierrez Del Arroyo, Michael A Temple + Show 1 more

Open Access

https://doi.org/10.3390/s22062111

Copy DOI

Abstract

Radio Frequency Fingerprinting (RFF) is often proposed as an authentication mechanism for wireless device security, but application of existing techniques in multi-channel scenarios is limited because prior models were created and evaluated using bursts from a single frequency channel without considering the effects of multi-channel operation. Our research evaluated the multi-channel performance of four single-channel models with increasing complexity, to include a simple discriminant analysis model and three neural networks. Performance characterization using the multi-class Matthews Correlation Coefficient (MCC) revealed that using frequency channels other than those used to train the models can lead to a deterioration in performance from MCC > 0.9 (excellent) down to MCC < 0.05 (random guess), indicating that single-channel models may not maintain performance across all channels used by the transmitter in realistic operation. We proposed a training data selection technique to create multi-channel models which outperform single-channel models, improving the cross-channel average MCC from 0.657 to 0.957 and achieving frequency channel-agnostic performance. When evaluated in the presence of noise, multi-channel discriminant analysis models showed reduced performance, but multi-channel neural networks maintained or surpassed single-channel neural network model performance, indicating additional robustness of multi-channel neural networks in the presence of noise.

Highlights

Physical-layer emitter identification, known as Radio Frequency Fingerprinting (RFF) or Specific Emitter Identification (SEI), is often proposed as a means to bolster communications security [1]
When the models were trained on data from Channel 7, they generalized well across a wide swath of channels (e.g., Channels 5–10), whereas when the models were trained on Channel 14, performance only roughly generalized to Channel 13 and only for MCA/Machine Learning (ML) and High-Capacity Convolutional Neural Network (HCCNN)
At low Signal-to-Noise Ratio (SNR) levels (SNR < 10 dB), multi-channel models for all four model types showed some advantage over single-channel models, though arguably, the performance in this region was already too weak (MCC 0.5) to be practical for RFF applications

Summary

Introduction

Physical-layer emitter identification, known as RFF or Specific Emitter Identification (SEI), is often proposed as a means to bolster communications security [1]. The underlying theory is that the manufacturing processes used for chip components create hardware imperfections that make each emitter unique, irrespective of brand, model, or serial number. Imperfections cause small distortions to the emissions of idealized signals, and those signal distortions can be learned by Machine Learning (ML) models to identify emitters solely from their emissions. This is useful for communications security applications where the reported bit-level identity (e.g., MAC Address, Serial Number) of a device cannot or should not be implicitly trusted. RFF can serve as a secondary out-of-band method for identity verification

Objectives

Methods

Results

Conclusion