Improved MIMO-SAR Echo Separation Scheme With Constrained/Generalized LASSO Regression: New Insights and Applications

The separation of multiple transmit waveforms with time and frequency synchronization constitutes a considerable challenge for multiple-input multiple-output (MIMO) synthetic aperture radar (SAR) systems. It is well-known that aliased signal returns may be separable by digital beamforming (DBF) on receive in elevation. However, the current orthogonal-waveform beamforming schemes significantly increase the hardware complexity. Moreover, the direction of arrival (DOA) mismatch issue caused by topographical variations significantly increases the complexity of the DBF process. To alleviate these issues, we here introduce a multiple-subpulse separation and weighting synthesis (MSS-WS) echo separation framework, which is formed using segmented phase coding (SPC) waveforms. The proposed MSS-WS scheme can halve the number of interferences from far arrival angles, allowing for a reduction of the system complexity. In addition, constrained/generalized least absolute shrinkage and selection operator (LASSO) regression is exploited to form the beamformer with relatively high robustness in terms of dealing with the presence of topographical variations. The so-called LASSO-based dynamic beam response (LASSO-DBR) technique introduced here contains two parts: the source localization and the beamforming based on the designed constraint matrices. In this respect, the proposed LASSO-DBR beamformer can produce a distortionless response to the desired signal and still yield wide nulls for the unwanted interferences. Using numerical simulations, we illustrate the feasibility and performance of the proposed MSS-WS framework using the LASSO-DBR beamforming technique.