For Full-Text PDF, please login, if you are a member of IEICE,|
or go to Pay Per View on menu list, if you are a nonmember of IEICE.
A Reduced-Complexity Signal Detection Scheme Employing ZF and K-Best Algorithms for OFDM/SDM
Takafumi FUJITA Atsushi OHTA Takeshi ONIZAWA Takatoshi SUGIYAMA
IEICE TRANSACTIONS on Communications
Publication Date: 2005/01/01
Print ISSN: 0916-8516
Type of Manuscript: Special Section PAPER (Special Section on Multi-carrier Signal Processing Techniques for Next Generation Mobile Communications--Part 1)
Category: Space Division Multiplexing
OFDM, SDM, MIMO, MLD, Zero-Forcing, K-best, M-algorithm, IEEE 802.11a,
Full Text: PDF(806.4KB)>>
This paper proposes a reduced-complexity signal detection scheme for Orthogonal Frequency Division Multiplexing with Space Division Multiplexing (OFDM/SDM) systems that utilize Zero-Forcing (ZF) and K-best algorithms. It is known that Maximum Likelihood Detection (MLD) with exhaustive search achieves mathematically optimal performance for SDM signal detection. However, it also suffers from exponential computational complexity against the number of transmit antennas and modulation order. In order to reduce the computational complexity of MLD, we apply the K-best algorithm for signal detection. It is known that the K-best algorithm itself inherently reduces the computational complexity of MLD because it avoids exhaustive search. In this paper, we propose the modified K-best algorithm, which exploits the ZF algorithm for initial symbol estimation. This initial symbol estimation improves the decoding accuracy of the original K-best algorithm. We evaluate the performance of the proposed scheme through computer simulations. The computer simulation results show that the performance degradation from the MLD algorithm is suppressed to just 1 dB or so in terms of the required Eb/N0 for packet error rate (PER) = 10-2, When either 16 Quadrature Amplitude Modulation (16QAM) or 64QAM is applied with three transmit and three receive antennas. In these cases, 87% and 99% fewer metric computations are required than the MLD algorithm. It is confirmed that the proposed MLD algorithm offers a significant reduction in the computational complexity from the MLD algorithm while suppressing the performance degradation.