Compressive Sensing (CS) provides a new paradigm in data acquisition and signal processing based on the assumption of sparsity and the incoherence of the received signal. Based on that concept, many radar front-end architectures have been studied with the implementation of CS. In these architectures, less data are collected but the radar scene can be recovered through CS with high probability. All these CS front-ends have always been stated as less complicated but never evaluated. The main motivation of this thesis is to find aspects of the complexity and performance which can be used for the characterization of these front-ends like the gain in signal-to-noise ratio (SNR), number of components, power consumption etc. In this thesis, we investigate three radar front-end architectures in CS. The first two are the Multi-coset (MC) Sampling and Analog-to-Information Converter (AIC) which are widely suggested for telecommunications and radar systems. The fourth front-end is novel as it contains metamaterial surface antenna elements. The performance and the complexity of each architecture are evaluated. The performance is compared with respect to the conventional reference case of a uniform linear array of antennas.