Wireless systems fundamentally rely on the characteristics of the physical propagation channel to ensure reliability, security, and sensing accuracy. However, a critical disparity exists between idealized theoretical models and the complex behaviors o...
Wireless systems fundamentally rely on the characteristics of the physical propagation channel to ensure reliability, security, and sensing accuracy. However, a critical disparity exists between idealized theoretical models and the complex behaviors of real-world channels. Practical environments introduce severe physical uncertainties, such as deep fading holes caused by indoor multipath interference, channel volatility driven by the high mobility of Low Earth Orbit (LEO) satellites, and off-grid leakage in advanced waveform modulation. This dissertation bridges the gap between theoretical design and physical reality by proposing a multimodal analysis framework to mitigate these uncertainties and ensure robust system performance.
First, we address the ``deep fading hole'' problem in indoor wireless sensing, where destructive multipath interference creates localized blind spots that compromise security. Unlike conventional fingerprinting methods that fail under abrupt signal attenuation, we propose SHARD, a Wi-Fi CSI-based secure region detection mechanism. SHARD introduces a hole elimination technique and a phase-distance metric to recover information from unreliable channel segments. Experimental results demonstrate a near-perfect true negative rate of 99.96\%, proving its robustness even in rigorous indoor multipath environments.
Second, we investigate physical layer security in Non-Terrestrial Networks (NTN), focusing on the challenges imposed by the high mobility of LEO satellite downlinks. The rapid channel fluctuations and open nature of satellite links create significant security vulnerabilities. We analyze an OTFS-based system assisted by a terrestrial relay and a cooperative jammer, designing a secrecy-aware jamming strategy that utilizes null-space beamforming. This approach suppresses interference at legitimate nodes without requiring the eavesdropper's Channel State Information (CSI). We derive closed-form expressions for the Secrecy Outage Probability (SOP) under composite Shadowed-Rician and Nakagami fading, verifying that the proposed strategy significantly enhances secrecy performance.
Third, we tackle the ``off-grid leakage'' problem in Advanced Waveforms, with a specific focus on Orthogonal Time Frequency Space (OTFS) modulation. In realistic time-varying channels, fractional delay-Doppler shifts cause severe energy dispersion. While multiple pilot structures are essential to mitigate the high Peak-to-Average Power Ratio (PAPR) of single-pilot schemes, they induce catastrophic ambiguity when standard patterns are used. We propose a pilot geometry optimization framework that minimizes the mutual coherence of the sensing matrix via a stochastic greedy search initialized with Costas arrays. Integrated with a guard-protected Orthogonal Matching Pursuit (OMP) estimator, this approach effectively eliminates structural interference and achieves robust channel estimation.
In conclusion, this dissertation demonstrates that mitigating physical channel uncertainties through rigorous modeling---ranging from geometric hole avoidance and secrecy analysis to optimized pilot structural design---is essential for next-generation wireless systems. The proposed mechanisms provide resilient and practical solutions for ensuring secure and reliable connectivity in the 6G and IoT eras.