A Two Stage Matching Pursuit Based Algorithm for DOA Estimation in Fast Time-Varying Environments

We propose a new directions of arrival (DOA) algorithm which provides very reliable estimates in fast time-varying environments. Popular DOA methods such as MUSIC and ESPRIT cannot be used in these environments because of their reliance on the estimation of a covariance matrix. The algorithm uses a two stage approach. The first stage uses a matching pursuit based algorithm to process each snapshot and determine DOA. These DOA estimates are aggregated at regular intervals and a subset of the DOA are selected and fed to a second processing stage. This stage consists of a bank of least squares estimators which are used to determine the final DOA estimate. The performance of the algorithm is evaluated for closely and widely spaced DOA over a range of SNR values.