| Design techniques for the partially adaptive space-time processor, motivated by a need for computational reduction, are proposed and evaluated for the interference cancellation as well as angle estimation in various operational environments. Based on the generalized sidelobe canceller (GSC), the power-space (PS) method for designing a fixed rank-reducing transformation is developed using the minimum mean-squared error (MMSE) design criterion. Another GSC-based method, the weighted-signal-subspace (WSS) method is developed to protect only the vulnerable parts of the quiescent main- beam pattern. These methods are unique because they approximate the global optimum using a non-iterative procedure and they provide superior performance relative to the previous design techniques. Because these methods produce fixed transformations, we call them "offline" methods. We also propose a hybrid design, which is based on the proposed offline methods and other online design methods, that is shown to be valuable because of its superior performance to the offline method and less computational complexity than the online method. The proposed design techniques are applied to various types of arrays including the uniform linear array, the uniform rectangular array, the circular ring array, and the nonuniform sparse array, for narrowband and broadband signals. A subarray design technique based on the offline methods is also proposed to reduce the hardware complexity as well as the computational complexity and demonstrated near optimal performance with great reduction in the number of beams and elements per beam. The performance of the designed partially adaptive arrays is evaluated with phase quantization error or random array errors. Two previous robustness techniques, the noise injection method and the iterative projection method, are combined with the offline-designed partially adaptive array to reduce the error effects in RF beamformers and eliminate the effects in digital beamformers. Finally, the existing generalized monopulse estimation (GME) technique is modified for application to the GSC and combined with the offline design methods to perform beamspace angle estimation in the presence of interference. Results show that the WSS method is preferred to the PS method in angle estimation and its performance is almost the same as the estimation of signal parameters via rotational invariance techniques (ESPRIT) and very close to the Cramer-Rao bound (CRB) at high signal-to-noise ratio (SNR). |
