Performance analysis of eigen-space based adaptive beamformer

An approach resulting in an approximate expression for the probability density function (pdf) of the normalized conditioned signal-to-interference-plus-noise ratio (NCSINR) for eigen-space based adaptive beamformer for narrow-band antenna arrays of arbitrary geometry is proposed in this paper. The analysis method is based on the first order perturbation expansion of the estimated eigenvalues, eigenvectors and projectors. The probability density is derived from the asymptotic properties of the sample covariance matrix of Gaussian signal model. It is shown that the pdf highly dependents on the power of desired signal in the training data. Computer simulation shows the correctness of the result.