Aspects of limited feedback, cooperation and coordination in multi-antenna cellular systems

Multi-antenna schemes have been shown to provide remarkable gains in terms of spectral efficiency and extensive research has been dedicated to studying those in detail, discovering new setups in which multi-input multioutput (MIMO) schemes can be used and making them practical. This thesis is concerned with certain aspects of two particular setups in which MIMO techniques may be implemented, which we summarize below. Limited feedback in the MIMO broadcast channel: We consider the downlink of a single cell where the base station (BS) has multiple antennas. A particular issue when dealing with such a system is that of channel state information at the transmitter (CSIT): this has been shown to play a cardinal role, particularly in the case when users are equipped with single antennas. Given the importance of acquiring that CSIT, and the fact that this acquisition comes at the cost of using resources on the uplink direction and introducing delays, a large body of literature has been concerned with limited feedback schemes for this setup. In this thesis, we contribute two ideas to try to make the most of the available feedback resource. Coordination and Cooperation in multicell systems: The above model of a single cell corresponds roughly to current cellular designs where frequency planning is used to separate cells that use the same frequency resources so that inter-cell interference is limited. This is however not very efficient and MIMO techniques may be used to allow for increased performance at full reuse. Thus incorporating more antennas at a given BS gives it the possibility to mitigate the interference it causes at users in other cells. This is one of the topics we deal with in this thesis and propose a scheme which, while requiring local CSI only, is shown to perform quite well. MIMO techniques in a multicell environment can moreover extend to multicell processing (MCP), whereby several BSs pool their antennas to essentially act as a giant MIMO transmitter. This, however, has significant costs in terms of backhaul for data and CSI sharing, which threatens the scalability of MCP. We start investigating how to deal with limitations related to both these aspects. i pa st el -0 00 06 16 2, v er si on 1 25 J un 2 01 0