On the Effects of UE Transceiver Non-Reciprocity in Coordinated TDD Multi-Cell MIMO Network

In this paper, we study the effects of effective channel non-reciprocity in coordinated TDD multi-cell MIMO network based on weighted sum rate (WSR) maximization. More specifically, we focus on UE transceiver non- reciprocity while the base stations (BS) are assumed to be perfectly calibrated, and both centralized and decentralized beamforming schemes are considered. In the centralized scheme, the cost function is constructed in a central controller using antenna specific UL pilots for channel estimation from all the connected BSs in the network. Then, even though the transceiver frequency response (FR) mismatches at the UE side corrupt the effective channel reciprocity, it is shown to have only a trivial impact on the WSR objective in such centralized case. However, when decentralized beamforming is deployed, the optimization is carried out in each BS and the corresponding cost function and optimization process depend on information acquired by over-the-air signaling between the BSs and all the users using precoded UL pilots. In this case, it is then shown that the transceiver FR mismatches at the UE side can cause severe performance degradation and even influence the convergence properties of the sum-rate optimization process. Further insight is provided for improving the performance by modifying the weight calculations in the optimization process and connecting users with good cell separations. Then a convergence-aware processing algorithm is also proposed to improve the performance of the decentralized scheme under UE transceiver non-reciprocity. Numerical experiments demonstrate that efficient processing algorithms for calibrating UE transceiver mismatches to be less than -30dB to -35dB are, in general, required in the decentralized system in order to achieve performance close to the ideal case without any RF imperfections.