Research On SAR Imaging Algorithm Based On New Computing Unit

Synthetic aperture radar(SAR)is a remote sensing means that actively emits electromagnetic waves,which is usually carried on a moving platform to achieve its function.Compared with traditional remote sensing methods,including optics and infrared schemes,SAR has the characteristics of full time period,full phenology,strong penetration and unconstrained antenna size.Nowadays,the constantly changing scenes and the rapid development of resolution ratio have made the signal processing of SAR face the increasingly severe challenges both in terms of computation load and throughput.Therefore,the hardware structure,which is the core computing module in the imaging process,is important for the applications of SAR.That is,the structure of the fast Fourier transform(FFT),has gradually become a new research hotspot.Thus,we carry out research work in this area.Firstly,the foundmental principle of SAR is introduced.The utilization of linear frequency modulation and matched filter accounts for the compromise between the range and the resolution of radar.Then,the simulation platform based on the vertical sidelooking strip-mode is implemented.The good imaging results shows that both the Range Doppler and the Chirp Scaling are capable of processing the echo data through the platform.In addtion,as the key module of SAR imaging process,the design of FFT unit is optimized at algorithmic and arithmetic level.Firstly,the statistics of computation during the Range Doppler and the Chirp Scaling algorithm testifies the importance of FFT.Secondly,for 2k-point fft,a new factorization method named "The Nearest Radices" which chooses two nearest radices in each step is put foward to minimize the amount of multiplication of Cooley-Tukey Algorithm.Finally,the unfolding method of butterfly processors is improved with the combination of factorization method and bit-serial units,which match the requirements of processing capacity and hardware efficiency satisfactorily.