Note: Your custom schedule will not be saved unless you create a new account or login to an existing account.
| Paper: | MLSP-P6.12 |
| Session: | Learning Theory and Models |
| Time: | Thursday, May 20, 15:30 - 17:30 |
| Presentation: |
Poster |
| Topic: |
Machine Learning for Signal Processing: Blind Signal Separation and Independent Component Analysis |
| Title: |
BLIND SOURCE SEPARATION WITH RANDOMIZED GRAM-SHMIDT ORTHOGONALIZATION FOR SHORT BURST SYSTEMS |
| Authors: |
Constantinos Papadias; Lucent Technologies | | |
| | Alexandr Kuzminskiy; Lucent Technologies | | |
| Abstract: |
A blind source separation problem for short burst systems is addressed by means of a constant modulus technique under orthogonal constraints. It is shown that a conventional Gram-Shmidt orthogonalization procedure normally exploited in similar applications may cause a non-uniform misadjustment distribution among the receiver outputs leading to an overal performance degradation. We propose a modified algorithm based on random reordering of the weight vectors before the orthogonalization stage and demonstrate its efficiency by means of simulations in a short burst MIMO environment. |
| |
| Back | |