The problem of bearing estimation of coherent signal impinging on an array of arbitrary geometry is studied. Both approaches are developed, the first one synthesizes the observed array data into the outputs of a linear uniform array and then processes them via conventional techniques such as spatial smoothing etc. The second approach is based on the invariant subspace rotation operation which is equivalent to the, translational displacement of the array, linearly independent signal vectors are obtained thereby to span perfectly the signal subspace. As compared with the conventional processing techniques, the approach based on invariant subspace rotation does not lead to an effective decrease in aperture size and therefore a decrease in resolution capability. The computer simulations are conducted to demonstrate the effectiveness of the presented approaches.