A Robust Eigenanalysis Interference Canceler Using Sensitivity-Based Number of Sources Estimation

Abstract

When the estimation of the number of sources is inaccurate and the expected signal is mixed in the training data, the conventional eigenanalysis interference canceler (EC) will result in the expected signal cancelation, which leads the performance of adaptive beamforming to be degraded significantly. In this paper, a robust eigenanalysis interference canceler (REC) algorithm is proposed to resolve this problem by using the minimum sensitivity-based number of sources estimation. In the proposed method, the number of sources is first pre-estimated by effectively using the minimum sensitivity function. Then, based on the eigenvalue decomposition of the array covariance matrix, the expected signal in the training data is determined and eliminated, so that the corresponding impact of the adaptive weight calculation can be avoided. Finally, the adaptive weight vector is obtained by orthogonally projecting the quiescent weight vector into the interference subspace. Simulation results show that the performance of the output signal-to-interference-plus-noise ratio (SINR) of the proposed algorithm has improved significantly when compared with the conventional adaptive beamforming methods.