The Role of Subspace Estimation in Array Signal Processing
Subspace-based algorithms for array signal processing typically begin with an eigenvalue decomposition of a sample covariance matrix. The eigenvectors are partitioned into two sets to get bases for signal and noise subspaces. However, the eigenvector subspace estimates are not the most accurate esti...
Saved in:
| Published in | 2019 53rd Asilomar Conference on Signals, Systems, and Computers pp. 1566 - 1572 |
|---|---|
| Main Author | |
| Format | Conference Proceeding |
| Language | English |
| Published |
IEEE
01.11.2019
|
| Online Access | Get full text |
| ISSN | 2576-2303 |
| DOI | 10.1109/IEEECONF44664.2019.9048994 |
Cover
| Summary: | Subspace-based algorithms for array signal processing typically begin with an eigenvalue decomposition of a sample covariance matrix. The eigenvectors are partitioned into two sets to get bases for signal and noise subspaces. However, the eigenvector subspace estimates are not the most accurate estimates obtainable from the data. Accuracy is defined here in terms of an intrinsic Cramer-Rao (CR) bound. A closed-form (non-iterative) algorithm that achieves the CR bound on subspace accuracy is derived, and examples are given for the applications of adaptive beamforming with a line array and DOA estimation with a planar array. The new algorithm requires far fewer snapshots (e.g. 10 to 100 times fewer) than the typical eigenvector approach to achieve a given level of performance. |
|---|---|
| ISSN: | 2576-2303 |
| DOI: | 10.1109/IEEECONF44664.2019.9048994 |