<div>Target detection, estimation, and discrimination have long been important research issues in the field of radar. Waveform design, analog signal processing, and digital signal processing are some techniques that can improve the detection, estimation, and discrimination ability. In this dissertation, we first address the sidelobe suppression from the waveform design point of view. We synthesize a non-constant modulus waveform for illumination of radar targets by applying a collection of constant modulus (linear frequency modulated (LFM) waveforms with different frequency offsets) waveforms from each transmitting array element in an antenna array, and we show from the ambiguity function that the non-constant modulus waveform has better performance with respect to the larger ambiguity function mainlobe-to-peak-sidelobe ratio than this ratio of a constant modulus (LFM-only) waveform. Furthermore, from the angular resolution point of view, the synthesized non-constant modulus waveform also has better performance than the angular resolution of a constant modulus waveform at the expense of the decrease in the signal energy on targets.</div><div><br></div><div>Secondly, we investigate radar delay-Doppler resolution enhancement from the digital signal processing viewpoint. We introduce the noise-target fringe analysis technique and combine it with the coherent CLEAN algorithm to provide accurate target parameter estimates in terms of delay, Doppler shift and intensity. Furthermore, the accuracy of target parameter estimates can be further improved by applying weighted non-linear least squares estimation.</div><div><br></div><div>Finally, we further aim for the improvement in radar delay-Doppler resolution. Instead of using the matched filter only, we propose a hybrid filter which combines a chirp matched filter and chirp mismatched filters. The hybrid filter output response shows much better performance in delay and Doppler resolution compared to the chirp matched filter output response. Thus, this hybrid filter design has better target identification capability than the original chirp matched filter. Furthermore, from a real implementation perspective, there is no need to significantly increase the hardware and software complexity of the radar, since we only need to mismatch the received waveform to another chirp waveform and perform some additional non-linear processing. Then a chirp radar system with high delay-Doppler resolution and accurate target discrimination ability can be easily achieved.</div>