Paul DE KERRET
ThesisCooperation with Distributed Feedback in Wireless Networks
Multiple-antenna based transmitter cooperation has been established as a promising tool towards avoiding, aligning, or shaping the interference resulting from aggressive spectral reuse. Although the impact of imperfect knowledge of the channel state information (CSI) is often investigated, it is usually assumed that the channel estimates are perfectly shared between all the transmitters. This assumption is however not adapted to many practical cases of transmitter cooperation between distant transmitters. Therefore, we focus in this thesis on the network scenario where the transmitters would like to cooperate in their transmission but can only imperfectly exchange on CSI which is acquired locally. This imperfect CSI sharing step gives rise to a CSI configuration, denoted as “distributed CSI”, where each transmitter has its own imperfect estimate of the global multi-user channel based on which it determines its transmit parameters.
We study first the impact of having distributed CSI over the precoder design. Specifically, we show that conventional precoding schemes are not adapted to the distributed CSI configuration and lead to poor performance. We then turn to another aspect of this CSI configuration which is to determine “Who needs to know what", when it comes to CSI at cooperating transmitters. In contrast to the resource-hungry solution consisting in providing the same CSI to all transmitters, it is shown how a non-uniform spatial allocation of the CSI to the transmitters can provide strong gains depending on the networks topology.