Pseudonymetry: Real-Time Interference Management System for Accountable and Cooperative Spectrum Sharing Using Embedded Watermarks

Le résumé fourni par la source

This dissertation presents the design, analysis, and experimental validation of Pseudonymetry, a real-time interference management framework that enables accountable and cooperative spectrum sharing. By embedding lightweight watermarks into secondary-user transmissions, the proposed approach allows devices to be aware of, identify, and respond to interference at the individual transmitter level. This makes it possible to achieve dynamic and safe coexistence between primary and secondary users (PUs & SUs) without requiring data demodulation or additional receiver infrastructure. In this way, the feedback-based mechanism helps close the gap between administrative spectrum coordination and real-time interference enforcement, enabling accountable and privacy-preserving coexistence among heterogeneous wireless systems. To achieve this objective, the dissertation introduces several watermarking techniques that are designed to operate under low-SNR and time-varying channel conditions. First, a pulse amplitude modulation (PAM) watermarking method is developed and theoretically analyzed to understand pseudonym bit error behavior and its impact on SU communication performance. Building on this, a coded pulse amplitude modulation (CPAM) scheme is proposed to improve robustness against channel fading and interference by encoding pseudonym bits as structured amplitude patterns across multiple symbols. In addition, a single-subcarrier CPAM variant is introduced to further minimize the impact of watermarking on the SU data link while preserving reliable pseudonym detection at the PU. This dissertation also presents a complete implementation of Pseudonymetry on the POWDER wireless testbed, including real-time pseudonym detection, database-driven feedback control, and machine learning–based interference classification. Over-the-air experiments demonstrate that interfering SUs can be stopped in less than 0.5 seconds for SNRs above -8 dB, and even at -10 dB within a few seconds. Furthermore, the results show that multiple interfering SUs can be simultaneously identified and stopped in less than two seconds, demonstrating the scalability of the Pseudonymetry system to detect and identify multiple simultaneous interferences at the PU. Moreover, field trials conducted at the Owens Valley Radio Observatory demonstrate that pseudonym-based interference detection is feasible in real-world radio astronomy environments, highlighting the potential of Pseudonymetry to complement existing dynamic spectrum sharing frameworks among heterogeneous wireless systems. In summary, this dissertation introduces Pseudonymetry as a scalable, efficient, and privacy-preserving solution for real-time interference management and accountable spectrum sharing using embedded watermarks. Through the development of new watermarking designs, theoretical analysis, and extensive over-the-air experiments and field validation, this dissertation demonstrates that Pseudonymetry enables effective real-time interference management while supporting cooperative and reliable coexistence among heterogeneous wireless systems.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.