Research On SAR/InSAR Real-Time Signal Processing Based On FPGA And ADSP

Synthetic aperture radar(SAR) can be operated during all-day and in all weather conditions. SAR can form two-dimensional(2-D) high-resolution images by using pulse compression in range direction and “synthetic aperture” in azimuth direction. Based on SAR, interferometric synthetic aperture radar(In SAR) system observes the same scene in slightly different angles to obtain two or more images with single-pass or multipass flying. By processing the interferometric phase of the two coherent complex images, the elevation information of the observed scene can be achieved, i.e., the digital elevation map(DEM) can be generated. In SAR plays an important role in the fields of geological monitoring, military reconnaissance, crop yield estimation, and resource exploration. Until now, SAR real-time processing has been developed widely, however, In SAR real-time processing has become a research focus recently.In this dissertation, we investigate the SAR/In SAR real-time signal processing. By using the collabrative work of FPGA and DSP, we achieve the real-time SAR/In SAR signal processing. SAR imaging algorithms are mainly carried out in FPGA, and In SAR processing algorithms are excuted in multiple DSP devices. According to the characteristics of single-pass In SAR system with short baseline, the SAR/In SAR real-time processing flowchart is firstly introduced in this dissertation, and the real-time processing results are given step by step. To guarentee the precision of In SAR real-time processing, the key factors affected the In SAR real-time processing are fully considered, including the non-ideal properties of signal transmission and echo acquisition link in SAR imaging, the POS equipment error on the accuracy of interferometric processing, and so on. This dissertation is cheifly organised as follows:Firstly, due to the non-ideal characteristics, such as AD conversion, DA conversion, analytic filters, in the hardware link, amplitude-frequency errors and phase-frequency errors can not be ignored. For the single-pass In SAR system, if the amplitude and phase properties of the two channels are different, the coherence of the In SAR image pairs will be deteriorated, then the interferometric phase error will be introduced and the accuracy of DEM generation will be degraded. Based on the result of the spectrum comparison method, the windowed frequency-domain is proposed to estimate the channel errors. Single channel amplitude and phase errors' estimation and correction are implemented by using FPGA, the detailed procedure is given and the experimental results verify the performance of real-time processing.Secondly, limited by the measured accuracy of POS equipments, residual motion error components still exist in the interferometric phase even after SAR motion compensation, which is caused by the baseline instability. Baseline error is a key factor which affects the accuracy of DEM. Even a small error will lead to a great impact on the In SAR phase and DEM generation. In the dissertation, the baseline error is estimated and corrected by using subaperture processing, therefore, the accuracy of measured DEM will be improved. For the real-time processing, we use FPGA and DSP to achieve the baseline error estimation and correction. Through the simulated In SAR data, we investigate the effectiveness of the method.Thirdly, we achieve the back-transmission and display of the real-time processing results through the Ethernet. Therefore, the real-time processing results can be monitored through a video dynamically. The detailed procedure, including the sending back of the real-time processing results, the integrating of partitioned data in different DSP given. Then the corresponding transporting interfaces are introduced separately, including the link ports data transmission between DSP and FPGA, the LVDS data transmission between the different FPGA, and the Ethernet data transmission between FPGA and PC. Based on the detailed real-time transmission process, the Signaltap sampling data are given to prove the validity of real-time transmission.