Optimized Gauss and Cholesky algorithms for using the LMMSE decoder in MIMO/BLAST systems with frequency-selective channels - Reduced-complexity equalization

The LMMSE (Linear Minimum Mean Square Error) algorithm is one of the best linear receivers for DSCDMA (Direct Sequence-Code Division Multiple Access). However, for the case of MIMO/BLAST (Multiple Input, Multiple Output/Bell Laboratories Layered Space Time), the perceived complexity of the LMMSE rec...

Full description

Bibliographic Details
Main Author: Silva, J. C. (author)
Other Authors: Souto, N. (author), Cercas, F. (author), Rodrigues, A. (author), Dinis, Rui (author), Jesus, S. M. (author)
Format: conferenceObject
Language:eng
Published: 2015
Online Access:http://hdl.handle.net/10400.1/7352
Country:Portugal
Oai:oai:sapientia.ualg.pt:10400.1/7352
Description
Summary:The LMMSE (Linear Minimum Mean Square Error) algorithm is one of the best linear receivers for DSCDMA (Direct Sequence-Code Division Multiple Access). However, for the case of MIMO/BLAST (Multiple Input, Multiple Output/Bell Laboratories Layered Space Time), the perceived complexity of the LMMSE receiver is taken as too big, and thus other types of receivers are employed, yielding worse results. In this paper, we investigate the complexity of the solution to the LMMSE and the Zero-Forcing (LMMSE without noise estimation) receiver's equations. It will be shown that the equation can be solved with optimized Gauss or Cholesky algorithms. Some of those solutions are very computationally efficient and thus, allow for the usage of the LMMSE in fully-loaded MIMO systems.