Design of Scalable Software Architecture for Spotlight SAR Imaging on UAV Platform

Synthetic aperture radar (SAR) for the Unmanned Aerial Vehicle (UAV) platform offers a wider range of applications for military and civilian purposes. Realization of real time SAR image formation software has always been a challenging task specifically when it operates to SPOT light mode. In case of UAV, where platform speed can go as low as $30 \mathrm{~m} / \mathrm{s}$, one of the main challenge becomes to process huge input IQ data size. In our scenario, this IQ data size can be maximum 13.37 GB. This paper focuses on the techniques to handle voluminous data size to realize full chain of $\omega-\mathrm{k}$ algorithm in real time. The proposed software architecture has been evaluated on Intel Xeon-D (VPX3482) DSP board. This board has 12 cores which provides up to 410GFlops computing performance. This board comes with 32GB of DDR4-2133 SDRAM on board memory. VxWorks 7.0 is used to bring in the real time performance of the software. Our results indicate that a full chain of $\omega$-k algorithm takes around 53 seconds for input data size of $13.37 \mathrm{~GB}(100703 * 34816)$, achieving real time performance.