Investigating SAMV Regarding its Suitability For FPGAs

SAMV is an iterative algorithm for determining the direction of arrival (DOA) usable for different sensor technologies, like Radar. It offers more benefits than conventional algorithms used for the same goal like MUSIC and ESPRIT. However, it is more complex and requires more computational performance. Since DOA algorithms are usually employed in embedded systems, dedicated accelerator cores are required for using SAMV in these applications. As there are no known investigations available that look into implementing SAMV as discrete hardware, this paper aims to provide an analysis of possible implementations. For this, different designs for an FPGA are proposed and studied. FPGAs are becoming more widespread in embedded systems and are often already integrated in common MPSoCs. That makes them a decent platform for very specialized accelerator cores. It was feasible to show an improvement in speed in range of three to seven times in comparison to a CPU implementation.