Fundamental limits and algorithms in decentralized and cooperative wireless networks

Network cooperation is known to bring multiplicative gains under certain ideal assumptions. However, current wireless settings cope with many challenging constraints, as tight delay constraints, fast-changing channels, or rate-limited backhaul links. The topic of analyzing how the non-fulfillment of the ideal hypotheses impacts the performance has generated great interest in the research community. Nevertheless, the main focus has been on settings in which the imperfect information is shared by all the nodes, which is not feasible in many scenarios. This thesis aims for shedding light on the performance of cooperative settings in which the information available at each node may be different. We focus on the distributed Network MIMO. This setting is characterized by two main aspects: The perfect sharing of the user's information data and the imperfect sharing of the channel information. We start by characterizing the Degrees-of-Freedom metric of the setting, which is an approximation of the capacity at high SNR. The contribution is twofold, as we provide both achievable schemes that considerably outperform the solutions in the literature and upper-bounds that illustrate up to which scale the distributed setting is harmed with respect to the perfect-sharing setting. The second perspective consists in restricting the transmission to the conventional paradigm of Zero-Forcing and analyzing the achievable rate at high SNR to understand whether the performance losses from decentralized information can be accurately calculated. We propose a novel zero-forcing scheme tailored to the decentralized configuration that asymptotically attains the centralized rate.