Date of Award

Spring 2019

Document Type


Degree Name

Doctor of Philosophy (PhD)


Electrical/Computer Engineering

Committee Director

Dimitrie C. Popescu

Committee Member

Chunsheng Xin

Committee Member

Linda Vahala

Committee Member

Rafael Landaeta


This dissertation explores the detection and false alarm rate performance of a novel transmit-waveform and receiver filter design algorithm as part of a larger Compressed Sensing (CS) based Multiple Input Multiple Output (MIMO) bistatic radar system amidst clutter. Transmit-waveforms and receiver filters were jointly designed using an algorithm that minimizes the mutual coherence of the combined transmit-waveform, target frequency response, and receiver filter matrix product as a design criterion. This work considered the Probability of Detection (P D) and Probability of False Alarm (P FA) curves relative to a detection threshold, τ th, Receiver Operating Characteristic (ROC), reconstruction error and mutual coherence measures for performance characterization of the design algorithm to detect both known and fluctuating targets and amidst realistic clutter and noise. Furthermore, this work paired the joint waveform-receiver filter design algorithm with multiple sparse reconstruction algorithms, including: Regularized Orthogonal Matching Pursuit (ROMP), Compressive Sampling Matching Pursuit (CoSaMP) and Complex Approximate Message Passing (CAMP) algorithms. It was found that the transmit-waveform and receiver filter design algorithm significantly outperforms statically designed, benchmark waveforms for the detection of both known and fluctuating extended targets across all tested sparse reconstruction algorithms. In particular, CoSaMP was specified to minimize the maximum allowable P FA of the CS radar system as compared to the baseline ROMP sparse reconstruction algorithm of previous work. However, while the designed waveforms do provide performance gains and CoSaMP affords a reduced peak false alarm rate as compared to the previous work, fluctuating target impulse responses and clutter severely hampered CS radar performance when either of these sparse reconstruction techniques were implemented. To improve detection rate and, by extension, ROC performance of the CS radar system under non-ideal conditions, this work implemented the CAMP sparse reconstruction algorithm in the CS radar system. It was found that detection rates vastly improve with the implementation of CAMP, especially in the case of fluctuating target impulse responses amidst clutter or at low receive signal to noise ratios (β n). Furthermore, where previous work considered a τ th=0, the implementation of a variable τ th in this work offered novel trade off between P D and P FA in radar design to the CS radar system. In the simulated radar scene it was found that τ th could be moderately increased retaining the same or similar P D while drastically improving P FA. This suggests that the selection and specification of the sparse reconstruction algorithm and corresponding τ th for this radar system is not trivial. Rather, a tradeoff was noted between P D and P FA based on the choice and parameters of the sparse reconstruction technique and detection threshold, highlighting an engineering trade-space in CS radar system design. Thus, in CS radar system design, the radar designer must carefully choose and specify the sparse reconstruction technique and appropriate detection threshold in addition to transmit-waveforms, receiver filters and building the dictionary of target impulse responses for detection in the radar scene.