Compact, cost-efficient and high-resolution imaging sensors are especially desirable in the field of short-range observation and surveillance. Such sensors are of great value in fields of security, rescue and medical applications. Systems can be formed for various practical purposes, such as detecting concealed weapons in public places, locating people inside buildings or beneath rubbles during emergency rescue, detecting landmine with small vehicle-based systems, and finding early-stage lesions inside human bodies. The advantage of such systems is that imaging can be achieved in real-time, which allows for safer and more effective operation as well as intelligence gathering. In order to observe moving targets, the system must operate quickly enough to generate a focused real-time 3-D image. Cost efficiency is another factor to consider for market reasons. Existing systems based on synthetic aperture radar (SAR) technology or array with sequential operation of identical transceiver pairs are costly, thus commercially difficult to be widely deployed. Their drawbacks are obvious when compared with compact, lightweight and high-resolution imaging systems which can provide real-time perception of the target. The combination of ultra-wideband (UWB) technology and real-aperture multiple-input multiple-output (MIMO) array offers a novel and practical solution for safe, reliable and cost-efficient imaging sensors with high resolution capability. UWB impulse with high fractional bandwidth controls the appearance of grating lobes caused by the sparsity of the array; while asymmetric transmit and receive arrays further reduce the number of required antenna elements within the array aperture. Compared to the conventional SAR systems, the combined advantages of UWB technology and MIMO technique can drastically reduce element density, data-acquisition time and the system costs. The development of UWB-MIMO array and related imaging algorithms is an important and novel research subject. The International Research Centre for Telecommunications and Radar (IRCTR) of Delft University of Technology has initiated several projects with the purpose to develop high performance array-based UWB imaging systems for different short-range applications. Under this framework, my research has achieved the following novelties and main results which are elaborated in the thesis: Properties of UWB focusing. A novel theory of ultra-wideband focusing is proposed, which has corrected the previous misunderstanding of the impact of bandwidth on imaging. The proposed theory shows that sparsity only ensures the existence of a grating lobe free region within the image space rather than the complete absence of grating lobes. Instead, the bandwidth relative to the center frequency, or fractional bandwidth, has deterministic impact on the formation of grating lobes. Specifically, the bandwidth of an imaging system must exceed its central frequency in order to benefit from the merit of wideband impulse properties, allowing high-resolution imaging with significantly reduced number of antenna elements. The proposed theory helps to resolve the dilemma between angular resolution and grating lobe level for a sparse aperture/array. MIMO array topology design. Novel design approaches for one-dimensional and two-dimensional MIMO arrays have been proposed. They allow direct formation of array topologies without the need of numerical optimization. The designed arrays following the proposed strategies are able to provide three-dimensional imaging capabilities and low grating/side lobe level with large element spacing and minimum number of antenna elements. Combination of UWB technology and MIMO array technique. It is demonstrated that the combination of UWB and MIMO array is a beneficial concept for microwave imaging systems. The MIMO array-based UWB imaging system is able to provide high down- and cross-range resolution, low side/grating lobe level, simplified RF scheme and significantly increased data acquisition speed. A procedure for designing UWB-MIMO array for short-range imaging has been formulated. The process translates requirements from the application into specific array specifications, such as operational frequency band, number of elements needed, aperture dimension, and array topology. Modified Kirchhoff migration for MIMO configuration. Modified Kirchhoff migration has been developed for both free-space and subsurface imaging. The modifications allow for extension of the classical Kirchhoff algorithm, which is based on the exploding reflector concept and is formulated for co-located transmit-receive antennas, to multistatic imaging with arbitrary positions of transmit and receive antennas. The algorithm combines the high quality imaging performance with acceptable computational costs, thus can be applied to demanding applications in practice. Range migration algorithm for MIMO configuration. Range migration algorithm is formulated for multistatic array configuration in the frequency-wavenumber domain for near-field imaging. By taking advantage of the computation efficiency of the fast Fourier transform (FFT), imaging speed of MIMO array-based system are significantly increased while maintaining high-quality imaging capabilities. Demonstration of UWB-MIMO system in different applications. Within the framework of several European projects, prototypes of MIMO array-based UWB imaging systems were designed and implemented for through-wall imaging (PROBANT), ground penetrating radar (Cadmium), and concealed weapon detection (RADIOTECT, ATOM). The feasibility of operational systems under various practical circumstances has been demonstrated. Thanks to the combination of UWB focusing, MIMO array and advanced digital beamforming algorithms, high resolution 3-D images were obtained using sparse aperture array with simplified front-end and low-cost equipments.