Design Of SAR Imaging And GMTI Algorithm Based On Multi-core DSP

The airborne Synthetic Aperture Radar(SAR)can not only perform high-resolution imaging of ground stationary objects at long distances,but also detect ground moving targets.All of these advantages makes it has great application value in the military and civilian fields.Because SAR signal processing has the characteristics of large amount of echo data,complex algorithm calculations,and high real-time requirements.And aircraft platform also has its characteristics.So how to design a high-performance SAR real-time signal processing system that suitable for aircraft is become one of the hot spots in current SAR research.The SAR real-time signal processing system designed in this thesis is based on a 3U-sized VPX bus architecture.This system integrates signal acquisition functions,signal processing functions,data storage functions and data playback functions.It consists of multiple functional modules,the core of which is a signal processing module.The signal processing module is designed based on the multi-core DSP chip TMS320C6678 and is responsible for most signal processing tasks.The TMS320C6678 chip integrates eight C66 x series cores with a working frequency up to 1.25 GHz.In addition,it has a three-level memory structure and rich interface resources.All of these advantages make it meet the requirements of high transmission bandwidth,large data storage and large calculation that required for SAR real-time signal processing.The main work of this thesis is to design embedded real-time signal processing software based on multi-core DSP chip TMS320C6678 to realize the functions of band SAR imaging,spotlight SAR imaging,GMTI processing and DBS imaging.Due to the SAR signal processing has the characteristics of large data volume and can be calculated in parallel,this thesis introduces a data segmentation storage method which divides the data into eight segments according to the distance.This method is convenient for multi-core DSP to process data in parallel.In addition,this thesis also proposes a method based on C6678 for transposing a non-square matrix with large amount of data in original address.In this method,the non-square matrix is divided into several small squares,and then the matrix is transposed for the small squares,so as to realize the transposition of the large squares matrix in original address.With this method we can save the storage space we used.At the same time,using this method can makes the transposed data more efficient in the subsequent DMA,which can improve the program execution efficiency.This thesis combines these two methods for software design.The specific design content includes software overall framework design and algorithm design.According to different modes,algorithm design can be divided into strip SAR imaging mode,spot SAR imaging mode and GMTI mode.In SAR mode,based on the Range-Doppler(RD)algorithm and the Polar Format Algorithm(PFA),this thesis uses the characteristics of the airborne platform and the measured data to adds corresponding algorithm modules to improve these two algorithm.The main improvement is to add a motion compensation module,a PGA estimation module,an azimuth modulation frequency estimation module and an image correction module.In addition,in the GMTI mode,this thesis combines GMTI processing with DBS imaging for engineering implementation.When designing algorithms in different modes,task assignment and data interaction scheme design between DSPs is performed according to the improved algorithm processing characteristics and project requirements.And then this thesis gives the algorithm processing flow in each mode and the specific design and implementation methods of each algorithm module.Finally,the radar real-time data is used to perform a functional test in each mode,and a clear ground scene image and moving target point trace can be obtained.After analyzing the image resolution and other indicators,they all meet the project requirements,which verifies the feasibility of the design of this thesis.