On the use of GPGPU computing for Space-Time Adaptive Processing on airborne systems

New generation radar systems offer both more frequency bandwidth and multi-channel receiving capability. This allows the use of advanced signal processing algorithms, which delivers a major performance leap, but the amount of data to be processed and the computational workload drastically increases. In the paper, we focus on Space-Time Adaptive Processing (STAP) which purpose is to remove ground echoes, in order to enhance slow moving target detection. The downside of STAP is its high computational cost. Graphical Processing Unit (GPU) processors are used for several years now to perform general purpose computing. Although not originally made for scientific computing, their high performance per watt efficiency on certain tasks makes it an ideal candidate for embedded radar systems where physical size and electrical consumption are very limited. We investigate the use of such processors for this application and compare their performances with classical general-purpose processor (CPU).