Subspace design of low-rank estimators for higher-order statistics |
| |
Authors: | Ivan Bradaric Athina P Petropulu Konstantinos I Diamantaras |
| |
Institution: | a Electrical and Computer Engineering Department, Drexel University, 3141 Chestnut Street, Philadelphia, PA 19104, USA b Department of Informatics, Technological Education Institute of Thessaloniki, 54101, Sindos, Greece |
| |
Abstract: | Higher-order statistics (HOS) are well known for their robustness to additive Gaussian noise and ability to preserve phase. HOS estimates, on the other hand, have been criticized for high complexity and the need for long data in order to maintain small variance. Since rank reduction offers a general principle for reduction of estimator variance and complexity, we consider the problem of designing low-rank estimators for HOS. We propose three methods for choosing the transformation matrix that reduces the mean-square error (MSE) associated with the low-rank HOS estimates. We also demonstrate the advantages of using low-rank third-order moment estimates for blind system estimation. Results indicate that the full rank MSE corresponding to some data length N can be attained by a low-rank estimator corresponding to a length significantly smaller than N. |
| |
Keywords: | Higher-order statistics Low-rank estimates Variance reduction System identification |
本文献已被 ScienceDirect 等数据库收录! |
|