Optimal designs for decentralized transmission with asymmetric channel state information

In order to capitalize on transmitter cooperation in a scalable and cost-effective manner, future generation networks are expected to decentralize an increasing number of operations which were originally conceived for centrally controlled systems. On the physical layer side, decentralizing the transmission opens a Pandora's box of research problems dealing with the possibly limited sharing of crucial control information, e.g., about the channel state. Starting from rigorous information theoretical models, the first part of this thesis extends known coding theorems for centralized systems to decentralized transmission with distributed CSIT, that is, by assuming that encoding is done on the basis of transmitter-specific channel state information. With this background at hand, we show that distributed precoding of Gaussian codewords achieves the capacity of a decentralized MIMO fading channel towards a single receiver, where the CSI is acquired through asymmetric feedback links. Surprisingly, we demonstrate that it may be suboptimal to send a number of data streams bounded by the number of transmit antennas as typically considered in a centralized setup. As a second main contribution, we then move to the problem of distributed precoding design for systems with multiple receivers. Building on the so-called theory of teams, we introduce a novel scheme, coined team minimum mean-square error (TMMSE) precoding, which rigorously generalizes classical centralized MMSE precoding to distributed CSIT. The proposed method is finally specialized to several examples of cell-free massive MIMO networks, leading to closed-form optimal solutions outperforming the available methods heuristically adapted from centralized transmission theory.