Comment on ´Efficient realisation of alternating projection algorithm for maximum likelihood direction finding´

For original paper see S.K. Oh and C.K. Un, ibid., vol.25, p.1325-6 (1989). By using the properties of the minimum norm spectral estimator, the original authors propose an efficient iterative ML algorithm that transforms quadratic forms in the AP algorithm into inner products of vectors for reducing the number of iterations to convergence by improving the initial estimates. The idea is very good. However, there are some problems associated with it. This comment discusses these problems.