A Detection Algorithm to Reduce the Condition Number of the Channel Matrix

Hyunwook YANG  Gyuyoung LEE  Seungwon CHOI  

Publication
IEICE TRANSACTIONS on Communications   Vol.E98-B   No.2   pp.280-287
Publication Date: 2015/02/01
Online ISSN: 1745-1345
Type of Manuscript: PAPER
Category: Fundamental Theories for Communications
Keyword: 
condition number,  projection,  lattice reduction,  Zero-Forcing,  

Full Text: PDF(1.8MB)
>>Buy this Article


Summary: 
When Zero-Forcing (ZF) is adopted as a detector, decreasing the condition number of the channel matrix increases the BER performance. In this paper, we propose a new detection algorithm which reduces the condition number of channel matrix down to nearly 2 on average. Since the least singular value of the channel matrix is a major factor determining the condition number, we, first, project the received signal into a space spanned by singular vectors that are orthogonal to the one corresponding to the least singular value. Then, LR decomposition is performed to reduce further the condition number of the projected channel matrix. Computer simulations show that the performance of the proposed algorithm is comparable to that of the ML detector for both correlated and uncorrelated channels. And also the proposed algorithm provides an at least 2dB improvement compared to the conventional LR-based Ordered Successive Interference Cancellation (LR-OSIC) detector with a Bit Error Rate (BER) of 10-3 and a comparable computation load.