Electronic Theses and Dissertations

Date of Award

5-1-2026

Document Type

Thesis

Degree Name

M.S. in Engineering Science

First Advisor

Mustafa Matalgah

Second Advisor

Hamid Bahrami

Third Advisor

Lei Cao

School

University of Mississippi

Relational Format

dissertation/thesis

Abstract

Accurate characterization of signal-to-noise ratio (SNR) is essential for reliable performance prediction in wireless communication systems, particularly in multipath-rich environments with significant channel variability. This thesis presents a measurement-driven analysis of empirical SNR behavior in 851 MHz non-line-of-sight (NLOS) cellular channels, based on a large dataset comprising nearly one million high-resolution measurements collected in a realistic indoor composite environment with the serving base station located outdoors. A dual-marker measurement methodology is employed to obtain consistent estimates of signal and noise power, enabling reliable SNR extraction over a wide dynamic range.

A statistical modeling framework is evaluated directly in the SNR domain to address limitations of conventional envelope-based approaches. Classical, generalized, and composite fading models are systematically compared using a bin-probability mean-squared-error criterion with enhanced sensitivity to reliability-critical tail regions. Among the candidate models, the ?–? distribution achieves the best agreement with the empirical data, accurately capturing the observed asymmetry and heavy-tailed behavior of SNR. Using the measured SNR dataset, key system-level performance metrics—including channel capacity, symbol error rate (SER), and outage probability—are evaluated for multiple M-PSK and M-QAM modulation schemes. The results reveal fundamental trade-offs between spectral efficiency and reliability and highlight discrepancies between idealized theoretical predictions and observed performance, particularly in low-SNR and deep-fading regimes that dominate system reliability.

Furthermore, analytical performance evaluation is conducted using the empirically fitted ?–? model and compared with a composite Generalized-K model. The ?–? model demonstrates closer agreement with measurement-driven results, confirming its effectiveness in representing complex composite real-world channel behavior.

Overall, this work establishes a unified measurement-driven framework integrating empirical characterization, SNR-domain statistical modeling, and measurement-based performance analysis with analytical validation. The results provide improved insight into real-world channel behavior and offer a practical basis for more accurate performance prediction and robust wireless system design.

Available for download on Wednesday, August 02, 2028

Share

COinS