Publications
Partially-adaptive LCMV Beamforming with Quadratic Pattern Constraints
Abstract
Linearly constrained minimum variance (LCMV) beamforming with quadratic pattern constraints (LCMV-QPC) is applied to reduced rank partially adaptive beamforming. The problem is formulated for general rank reducing transformations. Main beam and sidelobe pattern control is achieved by imposing a set of inequality constraints on the weighted mean-square error between the adaptive pattern and a desired beampattern over a set of angular regions. An iterative procedure for satisfying the constraints is developed that can be applied as post-processing to standard partially adaptive LCMV processors. The technique is applied to fixed beamspace, adaptive eigenspace, and hybrid eigenspace/beamspace beamformers for both a linear and non-linear array. The LCMV-QPC approaches provide additional control over the sidelobes at the expense of slightly lower SINR than standard LCMV methods, mainly due to loss in the depth of the nulls placed on the jammers. Among the reduced rank approaches, the hybrid eigenspace/beamspace technique achieves the best overall performance, which is close to full element-space LCMV-QPC performance.
This article may be downloaded for personal use only. Any other use requires the permission of the authors. The article (PDF) appeared in Proceedings of the 4th World Multiconf. on Systemics, Cybernetics and Informatics (SCI 2000), vol. VI, pp. 213-218, July 2000.