Neural Network Based Robust Adaptive Beamforming for Smart Antenna System

Abstract

As the growing demand for mobile communications is constantly increasing, the need for better coverage, improved capacity, and higher transmission quality rises. Thus, a more efficient use of the radio spectrum is required. A smart antenna system is capable of efficiently utilizing the radio spectrum and is a promise for an effective solution to the present wireless system problems while achieving reliable and robust high-speed, high-data-rate transmission. Smart antenna technology offer significantly improved solution to reduce interference level and improve system capacity. With this technology, each user’s signal is transmitted and received by the base station only in the direction of that particular user. Smart antenna technology attempts to address this problem via advanced signal processing technology called beamforming.
The adaptive algorithm used in the signal processing has a profound effect on the performance of a Smart Antenna system that is known to have resolution and interference rejection capability when array steering vector is precisely known. Adaptive beamforming is used for enhancing a desired signal while suppressing noise and interference at the output of an array of sensors. However the performance degradation of adaptive beamforming may become more pronounced than in an ideal case because some of underlying assumptions on environment, sources or sensor array can be violated and this may cause mismatch. There are several efficient approaches that provide an improved robustness against mismatch as like LSMI algorithm.
Neural network is a massively parallel distributed processor made up of simple processing units, which has a natural propensity for storing experimental knowledge and making it available for use. Neural network methods possess such advantages as general purpose nature, nonlinear property, passive parallelism, adaptive learning capability, generalization capability and fast convergence rates. Motivated by these inherent advantages of the neural network, in this thesis work, a robust adaptive beamforming algorithm using neural network is investigated which is effective in case of signal steering vector mismatch. This technique employs a three-layer radial basis function neural network (RBFNN), which treats the problem of computing the weights of an adaptive array antenna as a mapping problem. The robust adaptive beamforming algorithm using RBFNN, provides excellent robustness to signal steering vector mismatches, enhances the array system performance under non ideal conditions and makes the mean output array SINR (Signal-to-Interference-plus- Noise Ratio) consistently close to the optimal one.