Acceleration of advanced radar processing chain and adaptive pulse compression using GPGPU

Adaptive Pulse Compression (APC), which will provide super-resolution capability for current radar system, is an extremely time consuming process. It is always beneficial that have APC process been done faster to even real time to meet the demand of numerous applications, such as Airborne Sense and Avoid (ABSAA) system. In this paper, the supercomputing capability of Generalized-Purpose computing on Graphics Processing Units (GPGPU) is utilized to accelerate the APC algorithm of radar signal processing. Generalized APC algorithm, such as Matched Filter (MF), as well as advanced APC algorithms, such as Reiterative Minimum Mean-Square Error (RMMSE) and RMMSE based on Matched Filter outputs (MF-RMMSE), are implemented on Central Processing Unit (CPU) based platform along with efficient CPU based libraries, such as EIGEN and FFTW, and Graphics Processing Unit (GPU) based platform along with Compute Unified Device Architecture (CUDA) technology. Comparison of latency performance is provided between CPU and GPU based platforms. The results shows a great potential to expedite pulse compression and APC processing to meet the real-time requirement.