Date of Award

Spring 2015

Degree Type


Degree Name

Doctor of Philosophy (PhD)


Electrical and Computer Engineering

First Advisor

Mark R. Bell

Committee Chair

Mark R. Bell

Committee Member 1

Charles A. Bouman

Committee Member 2

David J. Love

Committee Member 3

Michael D. Zoltowski


Radar receivers typically employ matched filters designed to maximize signal to noise ratio (SNR) in a single target environment. In a multi-target environment, however, matched filter estimates of target environment often consist of spurious targets because of radar signal sidelobes. As a result, matched filters are not suitable for use in high resolution radars operating in multi-target environments. Assuming a point target model, we show that the radar problem can be formulated as a linear under-determined system with a sparse solution. This suggests that radar can be considered as a sparse signal recovery problem. However, it is shown that the "sensing" matrix obtained using common radar signals does not usually satisfy the mutual coherence condition. This implies that using recovery techniques available in compressed sensing literature may not result in the optimal solution. In this thesis, we focus on the greedy algorithm approach to solve the problem and show that it naturally yields a quantitative measure for radar resolution. In addition, we show that the limitations of the greedy algorithms can be attributed to the close relation between greedy matching pursuit algorithms and the matched filter. This suggests that improvements to the resolution capability of the greedy pursuit algorithms can be made by using a mismatched signal dictionary. In some cases, unlike the mismatched filter, the proposed mismatched pursuit algorithm is shown to offer improved resolution and stability without any noticeable difference in detection performance. Further improvements in resolution are proposed by using greedy algorithms in a radar system using multiple transmit waveforms. It is shown that while using the greedy algorithms together with linear channel combining can yield significant resolution improvement, a greedy approach using nonlinear channel combining also shows some promise. Finally, a forward-backward greedy algorithm is proposed for target environments comprising of point targets as well as extended targets.