Scholarlink

<p>Chipless Radio Frequency Identification (RFID) technology is rapidly emerging as a key enabler of intelligent Internet of Things (IoT) ecosystems, particularly within next-generation wireless edge networks empowered by artificial intelligence (AI) and 5G connectivity. Despite this potential, the realization of compact, high-capacity, and AI-adaptable chipless RFID systems remains challenging due to spectral congestion, strong inter-resonator coupling, and decoding sensitivity under realistic measurement conditions. To address these challenges, this paper presents an AI-enhanced 24-bit chipless RFID tag optimized for reliable operation within 5G-enabled wireless edge IoT frameworks. The proposed tag employs multiple T-shaped microstrip resonators patterned on a lossy RO4350B substrate, enabling dense spectral encoding within a compact footprint of 55 × 35&#xa0;mm<sup>2</sup>. Inter-resonator coupling is effectively mitigated through geometrical isolation and an optimized ground plane configuration, resulting in enhanced Q-factors and improved spectral purity. The electromagnetic performance of the tag is evaluated using full-wave simulations and bistatic radar cross section (RCS) measurements, including <i>φ</i>-scans (<i>θ</i> = 90°) and <i>θ</i>-scans (<i>φ</i> = 90°), to demonstrate angular stability and polarization tolerance. Mechanical flexibility tests are also conducted to verify the robustness of the resonance characteristics under bending conditions, confirming suitability for practical IoT deployments. For reliable identification at the wireless edge, a machine-learning-based decoding framework is implemented, integrating baseline stabilization, slot-locked notch detection, and probabilistic inference to mitigate noise, ripple, and spectral distortion. Experimental results show that the proposed AI-assisted reader achieves a bit-wise accuracy of 93% and a whole-code accuracy of 73%, significantly outperforming conventional threshold-based decoding methods. Overall, the synergistic integration of advanced RF hardware design and AI-driven edge analytics establishes an effective pathway toward energy-efficient, adaptive, and intelligent chipless RFID systems, supporting scalable and interoperable identification in future 5G-enabled wireless edge networks.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

AI-Enhanced Chipless RFID for Wireless Edge IoT: Machine-Learning Decoding and 24-Bit Spectrum Optimization

  • 期刊论文

  • 发表日期：

    2026年2月12日

查看全文
  • Kawther Mekki,
  • Syrine Neffati,
  • Nadia Ghezaiel,
  • Marwa Ben Slimene,
  • Hatem Rmili

摘要

Chipless Radio Frequency Identification (RFID) technology is rapidly emerging as a key enabler of intelligent Internet of Things (IoT) ecosystems, particularly within next-generation wireless edge networks empowered by artificial intelligence (AI) and 5G connectivity. Despite this potential, the realization of compact, high-capacity, and AI-adaptable chipless RFID systems remains challenging due to spectral congestion, strong inter-resonator coupling, and decoding sensitivity under realistic measurement conditions. To address these challenges, this paper presents an AI-enhanced 24-bit chipless RFID tag optimized for reliable operation within 5G-enabled wireless edge IoT frameworks. The proposed tag employs multiple T-shaped microstrip resonators patterned on a lossy RO4350B substrate, enabling dense spectral encoding within a compact footprint of 55 × 35 mm2. Inter-resonator coupling is effectively mitigated through geometrical isolation and an optimized ground plane configuration, resulting in enhanced Q-factors and improved spectral purity. The electromagnetic performance of the tag is evaluated using full-wave simulations and bistatic radar cross section (RCS) measurements, including φ-scans (θ = 90°) and θ-scans (φ = 90°), to demonstrate angular stability and polarization tolerance. Mechanical flexibility tests are also conducted to verify the robustness of the resonance characteristics under bending conditions, confirming suitability for practical IoT deployments. For reliable identification at the wireless edge, a machine-learning-based decoding framework is implemented, integrating baseline stabilization, slot-locked notch detection, and probabilistic inference to mitigate noise, ripple, and spectral distortion. Experimental results show that the proposed AI-assisted reader achieves a bit-wise accuracy of 93% and a whole-code accuracy of 73%, significantly outperforming conventional threshold-based decoding methods. Overall, the synergistic integration of advanced RF hardware design and AI-driven edge analytics establishes an effective pathway toward energy-efficient, adaptive, and intelligent chipless RFID systems, supporting scalable and interoperable identification in future 5G-enabled wireless edge networks.