Abstract: The pattern of each element in the conformal array has different orientations caused by the variable curvature of the conformal platform, which results in the polarization diversity of the conformal array. The polarization parameters of incident signals are considered in the array output data model. Based on the model of the cylindrical platform, the array output is extended by using the noncircular source property of the received signals, and the rank-reduction(RARE) estimator, subspace principle are adopted to estimate the direction-of-arrival(DOA). The proposed algorithm can achieve high-resolution without parameter matching. In the case of coherent sources, virtual spatial smoothing algorithm for decorrelating the signals is presented with the subarray of the extended virtual array. Simulation results verify that the proposed algorithm is effective in DOA estimation for noncircular signals based on cylindrical array with high resolution.