Subspace-based approach for DOA estimation using pilot symbol channel identification

In the paper, antenna processing is applied in the radio localisation frame. For a source of interest, the multipath directions of arrival (DOA) are estimated and the shortest/direct propagation path is identified. First, an unstructured estimate of the sampled channel impulse response is derived by use of pilot symbols. The channel response samples are then separately processed to recover the DOA of the relative paths. For stationary channels, it is suggested that smoothing be used in the case of a uniform and linear antenna array (ULA) to recover the source subspace. For fast fading channels, which is typically the case for high speed mobiles, it is shown that using a MUSIC like algorithm allows source subspace recovery by exploiting the gain diversity over a reduced number of slots with unchanged DOA and time delays. Separate processing of channel response samples reduces the constraint on antenna array size and allows comparison of the path lengths.