MIMO Radar DOD/DOA Estimation and Performance Analysis in the Presence of SIRP Clutter

i Abstract This thesis investigates the problem of target parameter estimation and performance analysis of multiple-input multiple-output (MIMO) radar in the presence of non-Gaussian clutter. During the past decades, multiple-input multiple-output (MIMO) radar has become a research subject of growing interest, due to its superior performance in many aspects over the traditional phased-array radar. Conventionally, MIMO radar clutter is modeled as Gaussiandistributed. This modeling, however, becomes unrealistic and inadequate in certain specific scenarios, where the clutter shows distinct non-Gaussianity. In the radar literature, one of the most notable and popular models for such non-Gaussian clutter is the so-called spherically invariant random process (SIRP) model. A SIRP is a complex, compound Gaussian process with random power and can be represented as the product of two components: a complex Gaussian process, called the speckle, and the square root of a positive scalar random process, called the texture. The goal of this thesis is to devise estimation algorithms for target parameters, more specifically, for direction-of-departures/arrivals (DODs/DOAs) of the targets, in a MIMO radar context in the presence of SIRP clutter, and to evaluate the ultimate performance of this estimation problem, in terms of performance bounds and of target resolvability. First, three DOD/DOA estimation algorithms are proposed, which differ from one another in the modeling of the texture, as well as in the respective likelihood functions that they are based on, but have in common that all three algorithms employ the same concept of the stepwise numerical concentration approach and thus have similar iterative procedures. Performance properties like convergence of iterations and computational complexity of the three proposed algorithms are then examined. Next, various Cramér-Rao-type bounds (CRTBs) for the DOD/DOA parameters in this context are derived for performance assessment and their relationships between one another are determined. The respective impacts of the texture parameters on the CRTBs are investigated to illustrate the effect of the clutter spikiness on the same. Then, the estimation performance achievable in the presence of SIRP clutter is studied from another point of view, namely, that of the target resolvability, which is quantified by the concept of the resolution limit (RL). As a result, an analytical, closed-form expression of the RL with respect to (w.r.t.) the angular parameters between twoThis thesis investigates the problem of target parameter estimation and performance analysis of multiple-input multiple-output (MIMO) radar in the presence of non-Gaussian clutter. During the past decades, multiple-input multiple-output (MIMO) radar has become a research subject of growing interest, due to its superior performance in many aspects over the traditional phased-array radar. Conventionally, MIMO radar clutter is modeled as Gaussiandistributed. This modeling, however, becomes unrealistic and inadequate in certain specific scenarios, where the clutter shows distinct non-Gaussianity. In the radar literature, one of the most notable and popular models for such non-Gaussian clutter is the so-called spherically invariant random process (SIRP) model. A SIRP is a complex, compound Gaussian process with random power and can be represented as the product of two components: a complex Gaussian process, called the speckle, and the square root of a positive scalar random process, called the texture. The goal of this thesis is to devise estimation algorithms for target parameters, more specifically, for direction-of-departures/arrivals (DODs/DOAs) of the targets, in a MIMO radar context in the presence of SIRP clutter, and to evaluate the ultimate performance of this estimation problem, in terms of performance bounds and of target resolvability. First, three DOD/DOA estimation algorithms are proposed, which differ from one another in the modeling of the texture, as well as in the respective likelihood functions that they are based on, but have in common that all three algorithms employ the same concept of the stepwise numerical concentration approach and thus have similar iterative procedures. Performance properties like convergence of iterations and computational complexity of the three proposed algorithms are then examined. Next, various Cramér-Rao-type bounds (CRTBs) for the DOD/DOA parameters in this context are derived for performance assessment and their relationships between one another are determined. The respective impacts of the texture parameters on the CRTBs are investigated to illustrate the effect of the clutter spikiness on the same. Then, the estimation performance achievable in the presence of SIRP clutter is studied from another point of view, namely, that of the target resolvability, which is quantified by the concept of the resolution limit (RL). As a result, an analytical, closed-form expression of the RL with respect to (w.r.t.) the angular parameters between two