Low complexity transceiver designs for MIMO AF relay system

Abstract

In this thesis, transceiver design methods for multipoint-to-multipoint (M2M) multi-input multi-output (MIMO) amplify-and-forward (AF) relay systems are suggested and analyzed. The transceiver is designed to minimize the mean-squared-error (MSE) while satisfying multiple power constraints at source and relay nodes. Due to the non-convexity, most previous methods have relied on numerical methods. We propose an analytical source filter and relay filter design methods to present closed-form solutions of sub-problems and obtain the global solution through an iterative algorithms. In the first chapter, we propose a transceiver design method for the case when source and relay nodes have sum-power constraints (SPC). In the second chapter, we propose a transceiver design for the situation when the antennas of source and relay nodes have per-antenna power constraints (PAPC). In the practical communication, each antenna has its own power amplifier, so this environment is more practical, but the problem has more complicated form. In the last chapter, we propose an antenna selection scheme to solve the cost and complexity of MIMO systems. Optimum antenna selection scheme is very inefficient because they indiscriminately need to be looked for in all cases. Through interpreting antenna selection in the MIMO systems to multiclass-classification, we propose a method of introducing machine learning technique into the field of antenna selection. We studied two antenna selection schemes in general MIMO system and extended massive MIMO system.