Adaptive beamforming for moving targets using Genetic Algorithms

Beamforming is a property of smart antennas that provides improvements on all sorts of wireless communications but it still requires studies. Generally, Least Means Squares (LMS) algorithm provides good adaptive beamforming and target tracking capability for an array of antennas since a reference signal is known. However, Genetic Algorithms (GA) have proven to perform very well in global optimization tasks, including blind DOA estimation. This works investigates the use of GA for beamforming on a CDMA environment under different Signal-to-Noise Ratios (SNR), considering a reference signal is known. Computational experiments on static targets have shown that, when compared to LMS, GA allow dramatically reducing the threshold SNR of performance for both correlated and uncorrelated interfering sources. On the other hand, LMS provided zero Root-Mean-Square-Errors (RMSE) at high SNR. Additionally, the performance has shown to be acceptable even for moving targets.