20 August 2019 in IEEE Signal Processing Letters
The multiple signal classification (MUSIC) algorithm is computationally expensive in the application to joint two-dimensional (2-D) direction-of-arrival (DOA) and time-of-arrival (TOA) estimation based on uniform circular array (UCA) using orthogonal frequency-division multiplexing (OFDM) signal. This letter proposed an efficient way to compute the 3-D spatial-temporal spectrum. We extended the manifold separation technique, by which we obtained the 3-D discrete Fourier transform (DFT) form of the spectrum. On this basis, we proposed two 2-D DOA and TOA estimators called FFT-MUSIC and Two-step FFT-MUSIC. The former computed the spectrum by large size fast Fourier transform (FFT) to reduce the grid error in searching DOAs and TOAs. The latter roughly located the DOAs and TOAs by relatively small size FFT, followed by the subspace-based technique for a fine-grained searching within a local grid. Simulation results showed that both estimators can reduce the computation cost by one to two orders of magnitude, as compared to their conventional counterparts, while maintaining a similar accuracy.
3 November 2017 in IEEE Communications Letters
A method for estimating the two-dimensional (2-D) direction- and time-of-arrival of orthogonal frequency division multiplexing (OFDM) signals that utilizes only three antennas is proposed. It firstly constructs an array manifold matrix not only based on the symmetry of circular geometry but also the time diversity of every OFDM subcarrier, so that the array aperture can be greatly increased. It then proposes that a virtual spatial smoothing method be used to partition the array manifold matrix into structurally identical sub-array manifold matrix, so that coherent multipaths can be correctly estimated by incorporating the eigenstructure-based techniques on the average of the sub-array covariance matrices. Simulation results are provided to demonstrate the effectiveness of the proposed method.