Search

Abstract: Current and future broadband cellular systems have to employ efficient techniques for the transmission and reception of high speed data. Equipping transmitters and receivers with multiple-antennas is a major step in this direction as it has the potential of providing a substantial spatial multiplexing gain. Unfortunately, interference from adjacent cells is an impediment to the spatial multiplexing gain promised by MIMO techniques. There exist solutions to mitigate the inter-cell interference in MIMO cellular systems, the most promising being coordinated multi-point (CoMP) transmission/reception (also known as network MIMO) and large-scale MIMO (also known as massive MIMO). The focus of this thesis is on multi-user MIMO techniques including precoding and user scheduling for large-scale and cooperative MIMO wireless systems. In this study, we design and analyze a near capacity-achieving non-linear precoding technique relying on vector perturbation (VP) along with a fair user scheduling algorithm for joint transmission network MIMO (usually operating in the frequency division duplex (FDD) mode). We consider practical conditions such as imperfect channel state information (CSI) due to the backhaul delay and per-base station (per-BS) power constraints. In addition, we propose an optimal VP technique minimizing the mean square error (MSE) of the received signal subject to per-BS power constraints. Although the array virtualization of network MIMO reduces the inter-cell interference to some extent (depending on the cluster size of coordinated BSs), the increase in transmit antenna array size is limited by the fading block length (coherence time of the radio channel). In the time division duplex (TDD) mode, the story is different thanks to the channel reciprocity. Massive MIMO or large-scale MIMO is a transmission/reception scheme for multi-cell MIMO, which works in the TDD mode and involves BSs, each with a large number of antennas, much larger than the number of users per cell. In this study, we design and analyze a non-linear precoding technique employing time domain VP (TDVP) for a large-scale (massive) MIMO system. To analyze the system we employ random matrix methods to avoid time-consuming Monte-Carlo simulations and get better insight into the problem. In addition, we propose a practical approach to mitigating pilot contamination for massive MIMO through a joint clustering and pilot reuse scheme. We propose pilot contamination precoding (PCP) as outer linear precoding prior to conventional precoding through a cooperative transmission scheme with three BSs involved in the coordination cluster.