Neural network-based adaptive beamforming for one- and two-dimensional antenna arrays

Abstract

In this letter, we present a neural network approach to the problem of finding the weights of one-(1-D) and two-dimensional (2-D) adaptive arrays. In modern cellular satellite mobile communications systems and in global positioning systems (GPS's), both desired and interfering signals change their directions continuously. Therefore, a fast tracking system is needed to constantly track the users and then adapt the radiation pattern of the antenna to direct multiple narrow beams to desired users and nulls interfering sources. In the approach suggested in this paper, the computation of the optimum weights is accomplished using three-layer radial basis function neural networks (RBFNN). The results obtained from this network are in excellent agreement with the Wiener solution.

Evaluation of a Neural-Network-Based Adaptive Beamforming Scheme With Magnitude-Only Constraints

“…This has motivated a growing interest toward implementations based on artificial neural networks (ANNs) [12], which are capable, via suitable training procedures, of accurately approximating complex nonlinear mappings, and admit very-largescale-integration hardware implementations. ANN-based algorithms have been successfully applied to several electromagnetics engineering optimization/synthesis problems (see, e.g., [13][14][15]) and, in particular, to adaptive beamforming schemes based on linearly-constrained MSE minimization [16,17].…”

“…In this paper, we propose an alternative scheme based on the MSE minimization which, unlike those in [16,17], involves magnitude-only (nonlinear) constraints on the array-factor in the desired-signal directions. As compared with the scheme in [16,17], for each constraint, we recover a real degree of freedom (arrayfactor phase) which we exploit in the optimization process for performance improvement.…”

“…As compared with the scheme in [16,17], for each constraint, we recover a real degree of freedom (arrayfactor phase) which we exploit in the optimization process for performance improvement. Via the Lagrange theory of constrained optimization [18], we reduce the problem to the unconstrained optimization of a non-quadratic functional (depending on the above auxiliary phase unknowns), for which we propose an iterative solution scheme.…”

“…Via the Lagrange theory of constrained optimization [18], we reduce the problem to the unconstrained optimization of a non-quadratic functional (depending on the above auxiliary phase unknowns), for which we propose an iterative solution scheme. Moreover, we address the ANN implementation, and assess the overall performance by comparison with standard schemes [16,17].…”

Approximate Kernel Orthogonalization for Antenna Array Processing

“…This feature decreases estimation errors or bit error rates. Neural networks [4] have been proposed for beamforming (e.g., [5]- [7]) and direction of arrival estimation (e.g., [8], [9]) among other array processing tasks. A com-A.…”