High Maneuvering Target Long-time Coherent Integration and Motion Parameters Estimation Based on Bayesian Compressive Sensing

Abstract

This paper intends to address the long-time coherent integration (LTCI) of high maneuvering targets in low signal-to-noise (SNR) environments with high computational efficiency and accuracy. The high-order motion parameters estimation, considering the acceleration and jerk, is modeled as the under-estimated linear regression and the complex-field Bayesian compressive sensing (BCS) algorithm is introduced to resolve the sparse recovery. To correct the high-order range migration (RM) and reduce the order of Doppler frequency migration (DFM), the adjacent cross correlation (ACCF) is applied. An effective method based on non-coherent integration is proposed to extract the self-term from ACCF result. Further, with the proper design of sensing matrix, the acceleration and jerk motion parameters are estimated by sparse reconstruction based on the complex-field BCS. Compared with the traditional methods based on time-frequency transform, like Lv's distribution (LVD), the proposed BCS algorithm is free from the interference of cross-terms and maintains the super-resolution ability, which provides better performance in parameters estimation and multi-targets discrimination. Finally, the numerical experiments validate the advantages of the proposed method in motion parameters estimation, super-resolution ability and computational efficiency.