Study of Class-Incremental Radio Frequency Fingerprint Recognition Without Storing Exemplars

TL;DR

The paper tackles scalable, privacy-conscious RF fingerprint recognition as devices continually join a network, making exemplar storage impractical. It introduces an exemplar-free class-incremental learning framework that freezes a pretrained RF feature extractor, adds per-task Adapters, and models per-class features with diagonal Gaussian Mixture Models to generate pseudo-features for rehearsal. A multi-teacher distillation scheme merges historical Adapters into a single inference Adapter, while time-domain masking augmentation boosts robustness under few-shot conditions. Experiments on large ADS-B datasets show improved average accuracy and reduced forgetting with low storage and no raw-data retention, supporting practical deployment in resource- and privacy-constrained environments.

Abstract

The rapid proliferation of wireless devices makes robust identity authentication essential. Radio Frequency Fingerprinting (RFF) exploits device-specific, hard-to-forge physical-layer impairments for identification, and is promising for IoT and unmanned systems. In practice, however, new devices continuously join deployed systems while per-class training data are limited. Conventional static training and naive replay of stored exemplars are impractical due to growing class cardinality, storage cost, and privacy concerns. We propose an exemplar-free class-incremental learning framework tailored to RFF recognition. Starting from a pretrained feature extractor, we freeze the backbone during incremental stages and train only a classifier together with lightweight Adapter modules that perform small task-specific feature adjustments. For each class we fit a diagonal Gaussian Mixture Model (GMM) to the backbone features and sample pseudo-features from these fitted distributions to rehearse past classes without storing raw signals. To improve robustness under few-shot conditions we introduce a time-domain random-masking augmentation and adopt a multi-teacher distillation scheme to compress stage-wise Adapters into a single inference Adapter, trading off accuracy and runtime efficiency. We evaluate the method on large, self-collected ADS-B datasets: the backbone is pretrained on 2,175 classes and incremental experiments are run on a disjoint set of 669 classes with multiple rounds and step sizes. Against several representative baselines, our approach consistently yields higher average accuracy and lower forgetting, while using substantially less storage and avoiding raw-data retention. The proposed pipeline is reproducible and provides a practical, low-storage solution for RFF deployment in resource- and privacy-constrained environments.

Figures (12)

• Figure 1: Overall framework
• Figure 2: Structure of Adapter
• Figure 3: Classifier structure
• Figure 4: t-SNE comparison of real and GMM-sampled pseudo-features
• Figure 5: Gaussian Mixture Model