Despite marked advances in hardware and signal-processing methodologies over the past fifty years, radar-based human physiological sensing lacks a unified benchmark of continuous-wave (CW) Doppler demodulation strategies under identical conditions. This dissertation implements and compares four time-domain methods, Arctangent Demodulation (AD), Extended Divide-and-Cross-Multiply (EDACM), Modified Divide-and-Cross-Multiply (MDACM), and Linear Demodulation (LD), and two spectral-domain techniques, Polyphase Basis Discrete Cosine Transform (PB-DCT) and Quadrature Cosine Transform (QCT), on human and mechanical mover datasets, evaluating micro-Doppler sensitivity, phase-wrapping immunity, computational burden, and SNR resilience. These results pave the way for modular, application-specific deployment of demodulation techniques. 

Extending this framework to non-sedentary activity detection, we benchmark hand-crafted time-series features and DCT spectrogram-based metrics with conventional classifiers and RNNs (LSTM, GRU), achieving substantial gains in accuracy metrics. Finally, we implement a commercial 28 GHz, 52-channel OFDM radar communication platform for dual-function operation of Joint Communication and Sensing, achieving vital-sign monitoring with no more than a 35 percent degradation in link-layer throughput. Together, these contributions establish a comprehensive benchmarking framework, advance non-contact activity classification, and demonstrate the feasibility of integrated radar communication platforms for future ubiquitous health monitoring.

Mohammad Shadman Ishrak is a Ph.D. candidate in the Department of Electrical Engineering at the University of Hawaiʻi at Mānoa, where he conducts his research under the supervision of Prof. Victor M. Lubecke. He holds a B.Sc. and M.Sc. in Electrical and Electronic Engineering from the University of Dhaka. As a Graduate Research Assistant, his work spans the design and implementation of continuous-wave and OFDM radar systems for non-contact vital-sign and activity monitoring, the development and benchmarking of novel demodulation algorithms, and the application of machine-learning techniques for sedentary versus non-sedentary classification. He has also collaborated with the Hawai‘i Institute of Marine Biology on low-frequency radar applications and mentors undergraduates and interns in radar-sensing capstone projects.