A study on the development of efficient algorithms for wireless MIMO relay communication systems

Abstract

This thesis proposes efficient algorithms for multiple-input multiple-output (MIMO) relay systems. MIMO relay systems have been extensively considered in wireless communications because of significant potential improvements in spectral efficiency and link reliability. The relaying strategies are, in general, classified into two categories according to the role at the relay node: amplify-and-forward (AF) relaying and decode-and-forward (DF) relaying. An AF relay, without having to decode the message signal, is easy to implement as compared to DF relay, and thus, AF relay is considered as an effective means of communication with low-cost and low-complexity. In addition, AF MIMO relay has attracted attention due to the potential merits of AF relay combined with multiple antennas. In this thesis, several efficient algorithms are developed for AF MIMO relay systems. In the first part, we investigate linear beamforming techniques for AF MIMO relay systems under two different types of spectrally efficient relaying protocols. In general, a half-duplex relay is preferable to full-duplex relay from an implementation aspect, but leads to a loss of spectral efficiency because the transmission and reception require orthogonal time resources. There are two main relaying protocols that overcome the loss of spectral efficiency: two-path relaying protocol and two-way relaying protocol. First, we propose the interference cancellation algorithms for alternate two-path relaying with multiple antennas. To cancel out the inter-relay interference, we exploit the orthogonal projection methods at the relay nodes or subtract an estimate of the interference at the destination node. As a result, the proposed algorithms effectively cancel out the interference with a very small loss in data rate as the number of antennas at the relay node increases. Next, we develop an iterative algorithm for designing an optimal linear precoder and decoder to minimize the sum of mean square...