Lattice-Reduction Aided Linear Equalization For Wireless Communications Over Fading Channels

Abstract

Modern wireless communications ask for high data rate, high transmission performance and low complexity. High data rate induces frequency-selective channels because of the relatively shorter symbol duration than the delay spread. Wireless links introduce fading which degrades performance and requires diversity techniques to combat. Orthogonal Frequency Division Multiplexing (OFDM) is an effective method to deal with frequency-selective channels since it facilitates low complexity equalization and decoding. To eliminate the effects of the channel nulls and fading, linear precoded OFDM is introduced to enable multipath diversity and the maximum likelihood decoder is used to collect the diversity. But, the low complexity provided by OFDM is sacrificed. Multi-antenna techniques are shown to be able to boost the data rate and also collect space diversity to combat fading. The V-BLAST (Vertical Bell Labs Layered Space-Time) scheme enables higher data rate than single-antenna setup does, but it also requires higher decoding complexity. As a combination, multi-input multi-output (MIMO-) OFDM has been widely studied to boost the transmit-rate and performance in terms of diversity. For MIMO-OFDM systems, many designs successfully exploit the joint space-multipath diversity when maximum likelihood (ML) detector is adopted at the receiver, which is well known for high complexity. To reduce the decoding complexity, linear equalizers are favored in practical systems, but they usually induce performance degradation. In this thesis, we first quantify the diversity of conventional linear equalizers for linear precoded OFDM, V-BLAST and MIMO-OFDM designs. Then, we propose lattice reduction (LR-) aided equalizers to improve the performance, and show that LR-aided linear equalizers achieve the same diversity order as that collected by ML detectors for (MIMO-) OFDM systems and V-BLAST systems. Simulation results corroborate the theoretical findings.