On Clutter Suppression Using Space-time Adaptive Processing

Clutter suppression is a primary concern in the application of moving target detection of moving-platform radars, e.g., airborne and spaceborne radar. The studies on this topic are largely covered by two parts. One is on optimum performance evaluation, which aims to aid in system design and performance comparison. The other focuses on adaptive processing, which is dedicated to make the adaptive performance approach its optimum counterpart. In this thesis, the clutter suppression using space-time adaptive processing (STAP) is studied, which as well concerns with the above two parts. Some significant results and conclusions are listed as follows.The expression of optimum performance in terms of the number of system degrees of freedom (NDoF) and clutter NDoF is derived, and a general criterion for performance evaluation is developed. The criterion is defined by the system and clutter NDoF, and is referred to as available space-time measurement (ASM) criterion. To estimate the system and clutter NDoF, the eigenvalue distribution of multi-dimensional bandlimited process is studied, and its approximate solution is derived. Accordingly, the NDoF estimation for various array geometries and system configurations, as well as taking into account non-ideal effects, is presented. A fast estimate on minimum detectable velocity (MDV) of the optimum processor in uniform sampling system is developed, which could facilitate the system design. In addition, the relationship among the MDV, resolution, and NDoF is clarified.In the second part, i.e., the adaptive processing, much attention has been paid to the knowledge-based clutter covariance matrix estimate as well as dimension- reduction. It is found that the clutter covariance estimate is equivalent to the reconstruction of the sample space of a random process. The NDoF of the to-be-reconstructed sample space decides the convergence of the estimate process, and the use of the a prior knowledge can effectively reduce the NDoF of sample space. The convergence of knowledge-based estimate method is theoretically analyzed in the case of knowledge mismatching, and a modified manipulation is proposed to mitigate the adverse effect of knowledge mismatch on convergence. On the other hand, some reduced-dimension (RD) methods that result in modest performance loss are investigated. It is found that such methods succeeded in preserving the system performance because they, indeed, introduced the knowledge of asymptote of the clutter eigenvectors into RD transform matrix construction. In this case, the RD matrix approximates to the optimum one that suffers no performance loss.Finally, the performance of multiple-input multiple-output (MIMO) based STAP is investigated. The signal and clutter model that is applicable to both code diversity and frequency diversity is established. And the system NDoF and clutter NDoF of both diversities is estimated in the case of ideally orthogonal transmit waveforms. According to the ASM criterion, the frequency diversity outperforms its code diverse counterpart and phased array counterpart in very sparse array systems with a remarkable margin. In addition, the extension of ASM criterion to the case of partially correlated waveform leads to a framework of performance evaluation for MIMO configuration.