Radar is an electromagnetic remote sensing system for the detection of reflecting targets. By transmitting pulses of very short duration, high resolution in the range direction can be easily achieved. However, traditional radar cannot distinguish between two side-by-side targets that are illuminated by the antenna beam at the same time. If we move the radar platform linearly while transmitting pulses in an oblique direction, the received echoes will become Doppler shifted. By using this information and processing the echoes coherently, we may achieve a very fine azimuth resolution, that is also range-independent. Therefore, with this technique, called synthetic-aperture radar (SAR), we may generate high-resolution 2D images of a target. With the progressive mainstreaming of software-defined radio (SDR) platforms, the usage of embedded computing devices for performing the required SAR processing from the transmitted signal is sought. However, the complexity of such algorithms makes this a challenging task. In this dissertation, we describe and implement an end-to-end SAR acquisition and processing system that targets such embedded architectures. In order to fulfill the imposed real-time requirements, the system makes use of a novel multilook SAR algorithm, and is programmed in order to make efficient use of the available resources. This system is to be used as part of the "Synthetic Aperture Radar using an Inflatable Antenna" (SARIA) project, that will fly in cycle 13 of the BEXUS program.