ASILOMAR 2020, Asilomar Conference on Signals, Systems, and Computers, 1-5 November 2020, Virtual Conference
Current cooperative transmission strategies for distributed MIMO systems are typically designed by assuming perfect, or at least perfectly shared, channel state information at the transmitters (CSIT). However, when this assumption is not met, the naïve application
of existing schemes encounters severe performance degradation. Recently, an information theoretical result unveiled the intriguing phenomenon that, for distributed MIMO systems with asymmetric CSIT, optimal encoding may require the transmission of a number
of data streams that goes beyond the classical upper bound predicted by the number of antennas in the system. In this work, we explore the implications of the aforementioned result from precoder design point of view. More precisely, we propose a method for distributed precoding design which optimally minimizes the expected interference.
As in the information theoretical result, the key idea lies in allowing the transmission of additional data streams. We show that this trick helps in transforming an otherwise difficult robust optimization problem into a convex formulation. Numerical simulations
corroborate the robustness gain in terms of achievable sum-rate of the proposed method over traditional schemes.