A real-time SAR imaging system based on CPUGPU heterogeneous platform

Abstract

With the features of exceptional computing power, low-cost, low power consumption characteristics, the CPU-GPU heterogeneous platform provides an alternative choice for high performance parallel computing. This paper introduces an efficient SAR imaging system design and implementation, which is based on such heterogeneous platform. Tested with Envisat ASAR raw data, the system meets the demands of real-time SAR processing. Furthermore, GPU-based imaging results in a 13 times speed up than multithreaded CPU implementation.