On the role of raw data oversampling in signum coded SAR imaging

Synthetic aperture radar (SAR) image formation requires a proper processing of the raw data, that consists essentially of a 2D space-variant convolution between the received signal (raw data) and the system point spread signal (PSS). Due to the large amount of data to be processed, the received signal is quantized using 3-8 bits, properly coded and transmitted to the ground, where the image formation process is performed. An alternative processing technique is based on coding at only one bit the involved signals, so that only the information about the sign of the signals is preserved. This processing technique is based on the result that the image obtained by considering only the signum of the signals to be processed is approximately equal, but for a scaling factor, to the image obtained starting from the conventionally quantized signals. Also the image phase exhibits acceptable quality. The main advantages of this processing method are the drastic transmission data rate reduction. The discrepancies between the images obtained by processing conventionally quantized data and one bit coded data are mainly due to the aliasing effects between higher order harmonics generated by the SC operation, that is inherently nonlinear and generates harmonics of the input signal. By increasing the sampling frequency this effect can be reduced, so that the quality of obtained images can be further improved. This is coherent with theoretical results regarding the reconstruction of a band limited signal by its zero crossing: high quality reconstruction is possible as long as the phase is highly random and the sampling rate sufficiently high