Modification of Multichannel Reconstruction Algorithm on the SAR With Linear Variation of PRI

This paper focuses on the signal processing for synthetic aperture radar (SAR) with fast linear variation of PRI which was introduced to settle the blind range problem in conventional SAR with constant PRI. Since periodic pulse loss is resulted from the variation of PRI, the distribution of the completely received pulses strongly deviates from the ideal uniform one. To settle the problem of nonuniform sampling, multichannel reconstruction algorithm is employed. As the reconstruction requires a priori knowledge of the precise distribution of the samples, this paper first derives the position of the lost pulses. Further, in order to reduce the nonuniformity, an optimal PRI variation scheme is designed. Moreover, considering the drastic degradation of the AASR and the SNR caused by the conventional multichannel reconstruction, we propose a modification where the number of processed frequency bands is reduced, and then the increased degrees of freedom (DOF) are exploited to minimize the power of the residual ambiguities and the noise via a linear constraint minimum power (LCMP) method. Compared to the conventional reconstruction, the modified one dramatically improves the ISLR, the AASR, and the SNR, which is validated by simulation experiments.