Quantization noise analysis of fixed-point CS algorithm for SAR

Quantization noise from fixed-point processing and limited word length storage affect real-time imaging and system miniaturization in spaceborne synthetic aperture radar (SAR) imaging processors, A quantization error model was developed for fixed-point SAR processing using the well-known chirp scaling (CS) algorithm. The expression of noise-to-signal (NSR), system word length, and FFT length was deduced. Test images from the raw Radarsat-I data were then analyzed using the fixed-point CS algorithm with various system word lengths. The image qualities are consistent with theoretical values. The analysis shows that the key parameters, such as the system word length, of fixed-point SAR imaging processors may be determined by calculating the total NSR of the system outputs.