Subspace-Based Method for Direction Estimation of Coherent Signals with Arbitrary Linear Array

In this paper, we propose an interpolated computationally efficient subspace-based method without eigendecomposition (ISUMWE) for the direction-of-arrivals (DOA) estimation of narrowband coherent signals in arbitrary linear arrays. ISUMWE estimates DOA based on the outputs of the virtual array by using an interpolation transform technique. Therefore, it overcomes the common restriction of uniform linear array (ULA) geometry when estimating coherent signals and becomes suitable for more general array geometry than ordinary methods. Meanwhile, the coherency of incident signals is decorrelated through a linear operation of a matrix formed from the cross-correlations between some sensor data in a designed virtual array which can be computed from the linear transformation of sensor data in the real array, where the effect of additive noise is eliminated. Consequently, the DOA can be estimated without performing eigendecomposition, and the noise pre-whitening which is required in traditional interpolation procedure can be avoided. As a result, the ISUMWE extends the application of the original subspace-based method without eigendecomposition (SUMWE) into arbitrary linear array with high accuracy and low computational complexity. The numerical results demonstrate the validity of the proposed method.