Design and Optimization of Distributed Multiuser Cooperative Wireless Networks

Employing multiple antennas at the transmitter and receiver sides has been identified as the key enabler for high spectral efficiency in point-to-point communication, since it facilitates multiplexing of several data streams in space rather than in time/frequency. In this work, we are interested in achieving spatial multiplexing (SM) through efficient cooperative relaying schemes, even if both the transmit and receive antennas are distributed. To this end, we focus on wireless multiuser networks and aim at designing novel cooperative communication protocols, developing corresponding transmission and signal processing techniques, and optimizing the network performance. This thesis is presented in two main parts. In the first part, we consider coherent multiuser amplify-and-forward (AF) relaying, where a set of source-destination (S-D) terminal pairs communicate concurrently over the same physical channel and a set of AF relay nodes assist the communication in a half-duplex scheme. It is known that multiuser interference can be cancelled with sufficient spatial degrees of freedom, i.e. relays. When additional relays are introduced to the network, we show that further multipleinput multiple-output (MIMO) gains can be achieved in a distributed manner through efficient (complex) relay gain optimizations. We choose two objective functions to maximize: sum rate and minimum link rate (max-min fairness). Moreover, it is shown that distributed diversity can be attained at the destinations through either relay selection or max-min type relay gain optimizations. Such gain allocation schemes require global knowledge of the channel coefficients between all participating nodes, which in practice may well diminish the spatial multiplexing gain. Addressing this issue, we introduce a new distributed gradient based gain allocation scheme, which substantially reduces this overhead. Key to this is the proof, that the gradient of the destination signal-tointerference-plus-noise ratio can be calculated in a distributed manner based on local channel state information (CSI) at the relays and limited feedback from the destinations. The two sine qua non assumptions for the efficiency of coherent multiuser relaying are perfect CSI knowledge per relay and globally phase-synchronous relays. We take a