Implementing radar algorithms on CUDA hardware

This paper investigates the implementation of radar algorithms on Graphics Processing Units (GPUs). The focus is on electronically scanned search radars. GPUs enable to develop high performance digital processing systems with limited development time. It is possible to employ a single commercial board (an NVIDIA GeForce 680 GTX in our case) to perform all the algorithms of a search radar: downconversion, amplitude/phase correction, pulse compression, beam forming, spectrum analysis, and CFAR noise floor estimation.