Wideband direction-of-arrival estimation using fast chirplet-based adaptive signal decomposition algorithm

The direction-of-arrival (DOA) is one of important parameters, in addition to time-of-arrival (TOA) and attenuation, to describe most radio propagation channel models. Time-frequency analysis is a powerful technique to deal with time-variant or non-stationary signals. Although M.G. Amin (see Gersham, A.B. and Amin, M.G., IEEE Sig. Proc. Letters, vol.7, no.6, p.152-5, 2000; Belouchrani, A. and Amin, M.G., IEEE Sig. Proc. Letters, vol.6, no.5, p.109-10, 1999) and L. Jin et al. (see IEEE Int. Symp. on Circuits and Systems, p.375-8, 2000) combined it with a coherent signal subspace (CSS) approach, it is not easy to extend from the narrowband to the wideband case. We first overview the fast chirplet-based signal decomposition algorithm, then build the general time-frequency covariance matrix to overlap the narrowband and wideband of CSS. After discussing the difference between the two cases, we propose a wideband DOA estimation algorithm with fast chirplet-based adaptive signal decomposition. The algorithm has no restriction to the array manifold. We give the comparison of our proposed algorithm with the optimal solution of maximum likelihood (ML). We can see that the proposed algorithm has very low complexity. Its robust performance is demonstrated with numerical simulations