Fixed-point simulation technology for SAR real-time imaging system

In this paper, a fast fixed-point simulation technology is proposed for a multi-channel synthetic aperture radar (SAR) imaging system. The overall computations in the SAR imaging algorithm is explicitly analysed. FFT operations account for most memory consumption in the imaging flow. Accordingly, a fixed-point FFT algorithm is adopted due to the trade-off between precision and memory occupation of the FPGA implementation. Furthermore, this paper describes an optimized word-length configuration method based on theoretical analysis of architecture of Radix-22 FFT. This methodology is verified in a SystemC development environment due to its superiority of fast and bit-accurate system-level simulation capability. Finally, the proposed SystemC based fixed-point imaging algorithm and the Matlab based floating-point imaging algorithm are both conducted to process point array target raw data. Peak side-lobe ratio (PSLR) and integrated side-lobe ratio (ISLR) is evaluated. Experimental results indicate that profiles in the two cases do not differ greatly, which validates that the proposed fixedpoint algorithm with less resources occupation has comparable accuracy and can meet system requirements.