2019/07/20 by Jiabao Yu, Yu, Jiabao, Aiqun Hu +11 · 2 citations
Computer Science · Engineering · #Cryptography and Security (cs.CR) #Digital Media Forensic Detection #FOS: Computer and information sciences #FOS: Electrical engineering #Signal Processing (eess.SP) #Terahertz technology and applications #Wireless Signal Modulation Classification #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.1907.08809
openalex publication_date 2019/07/20 · openalex created_date 2020/07/16 · openalex updated_date 2026/07/28
Radio Frequency Fingerprinting (RFF) is one of the promising passive\nauthentication approaches for improving the security of the Internet of Things\n(IoT). However, with the proliferation of low-power IoT devices, it becomes\nimperative to improve the identification accuracy at low SNR scenarios. To\naddress this problem, this paper proposes a general Denoising AutoEncoder\n(DAE)-based model for deep learning RFF techniques. Besides, a partially\nstacking method is designed to appropriately combine the semi-steady and\nsteady-state RFFs of ZigBee devices. The proposed Partially Stacking-based\nConvolutional DAE (PSC-DAE) aims at reconstructing a high-SNR signal as well as\ndevice identification. Experimental results demonstrate that compared to\nConvolutional Neural Network (CNN), PSCDAE can improve the identification\naccuracy by 14% to 23.5% at low SNRs (from -10 dB to 5 dB) under Additive White\nGaussian Noise (AWGN) corrupted channels. Even at SNR = 10 dB, the\nidentification accuracy is as high as 97.5%.\n