A persymmetric modified-SMI algorithm
Signal Processing
Performance analysis of fast projections of the Hung-Turner type for adaptive beamforming
Signal Processing - Special issue on subspace methods, part I: array signal processing and subspace computations
Adaptive filter theory (3rd ed.)
Adaptive filter theory (3rd ed.)
A model selection rule for sinusoids in white Gaussian noise
IEEE Transactions on Signal Processing
IEEE Transactions on Signal Processing
Computationally efficient maximum likelihood estimation ofstructured covariance matrices
IEEE Transactions on Signal Processing
IEEE Transactions on Signal Processing
| Hi-index | 0.00 |
We consider detection-estimation of Gaussian sources in coloured Gaussian noise for scenarios where a training data set is provided in addition to the primary data set that may contain source signals of interest. We propose a generalised likelihood-ratio test technique based on the optimisation of the likelihood ratio (LR) function that involves both data sets. This optimisation problem is non-convex and so requires some assessment of the quality of its results. The proposed assessment is based on the previously introduced scenario-free lower bound for maximum LR. Joint optimum processing of the two data sets is shown in general to be different from the conventional adaptive technique, whereby the training data set is separately processed, then such estimates are used for primary data processing. We demonstrate that beyond certain threshold conditions, our technique provides an estimation accuracy that is consistent with the corresponding Cramér-Rao bound, whereas maximum likelihood "performance breakdown" is found to occur for scenarios not satisfying such conditions.