HyDRA network fuses CNN, Transformer, and Mamba for advanced RFFI

Researchers have developed HyDRA, a novel hybrid dual-mode network designed for Radio Frequency Fingerprint Identification (RFFI). This architecture integrates an optimized Variational Mode Decomposition (VMD) with a combination of Convolutional Neural Networks (CNNs), Transformers, and Mamba components. HyDRA is capable of supporting both closed-set and open-set classification tasks, demonstrating state-of-the-art accuracy in closed-set scenarios and robust performance in identifying unauthorized devices. The system has been deployed on an NVIDIA Jetson Xavier NX, achieving millisecond-level inference speeds with low power consumption, making it suitable for real-time wireless authentication. AI

IMPACT This research advances RFFI capabilities by integrating multiple advanced AI architectures for improved wireless security.

RANK_REASON The cluster describes a new research paper detailing a novel network architecture for RFFI. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• convolutional neural network
• Hanwen Liu
• HyDRA
• Mamba
• Mamba Linear Flow Encoder
• NVIDIA Jetson Xavier NX
• Transformer Dynamic Sequence Encoder
• transformers
• Variational mode decomposition method for estimation of GNSS data quality from a smartphone

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →