A novel method in compensating random perturbations in eigen-based subspace method

The problem of random perturbations as induced by the nonstationary nature of the emitting sources and the sensing environment in direction-of-arrival (DOA) determination by sensor array is considered. Focus has been made on the effect of these random perturbations on the covariance matrix as generated by the sensor data. Based on certain assumptions, we show that the perturbed covariance matrix can be expressed as the Schur-Hadamard product between the ideal one and a perturbation matrix which is independent of both the DOAs and the source covariance. However, unlike the MLE and exhaustive search approach used to determine the DOAs, we propose a fast, simple and robust method to compensate for the random perturbations. Simulations on MUSIC with different models of perturbation distributions are carried out and promising results are obtained