The Realization Of Real-time Imaging Algorithm For FMCW SAR

The frequency modulated continuous wave synthetic aperture radar (FMCW SAR)has the salient characterics of light weight and high resolution, which will have a broadprospect particularly in the military applications. The works presented in this paperwere carried out based on the parameters of a specific radar. In this paper, the real-timeimaging technique of miniature SAR is researched in depth, the imaging algorithm ofFMCW SAR is analyzed and picked, also the real-time requirements and key techniquesof imaging system are demonstrated in detail. The key issues of massive matrixtransposition and multi-core parallel processing are solved, so that the imagingalgorithm can be realized on the multi-core DSP effectively and also the rigid real-timerequirement is fulfilled, which established the groundwork for the real-time imagingminiature SAR system based on TMS320C6678.As we all know, the SAR real-time processing system must have the ability ofquickly matrix transposition. Conventionally, a transposition board is applied forperforming the massive matrix transposition operation. However, this method not onlyincreases the cost but also leads to bulky size and high power consumption, so thisstrategy is not suitable for the small size unmanned aerial vehicles (UAV), which haslimited load space. To resolve this problem, a novel method is proposed. In theproposed algorithm, the latest highest-performance multicore DSP-TMS320C6678isused as the kernel device, and the DDR3SDRAM is used to buffer the massive matrixdata which is transmitted in EDMA3(enhanced direct memory access version3) way.And the matrix transposition is realized on the evaluation module(EVM) board ofTMS320C6678. The proposed method has the characteristics of simple realization andhigh transmission efficiency. Besides, the proposed does not need any extra processingboard.In the multicore development and application aspect, the echo simulation data ofsingle-point target and multi-point targets are processed correctly based on the singalcore of TMS320C6678. Moreover, based on the singal-core SAR imaging, themulti-core parallel imaging is realized. Furthermore, by distributing imaging tasksreasonablly, reducing the spending of serial tasks, increasing the degree of systemparallelism and setting the correct parallel data transmission parameters of EDMA3, theefficiency of multi-core parallel processing is greatly improved. And the parallelimaging is verified and tested on EVM board. Finally, a deep analysis on the real-timeimaging requirement for the specific radar is carried out in detail.