Errors in signal subspace methods in the presence of moving sources

The effects of angular motion on signal subspace methods, in particular the MUSIC algorithm, are investigated for a finite number of unweighted data samples. The perturbation of the signal subspace due to source motion is found, and the associated angle of arrival estimate is calculated in terms of the array manifold and the received signals. Assuming the received signal samples to be random circular Gaussian, expressions for the mean and variance of the arrival angle estimates are given in a noiseless case, and compared with those due to noise alone. For a specific scenario, there is a limit on the number of snapshots for which these expressions are accurate. However, they are useful in that they allow an insight into the factors determining motion induced errors.<
>