Robust MSE-based transceiver optimization for downlink cellular interference alignment

In this paper, we consider mean squared error (MSE)-based robust transceiver optimization for multi-user multi-input multi-out (MU-MIMO) cellular network that exploits interference alignment (IA) for its downlink communication. First, we consider the conventional sum-MSE minimization problem, where the MSE is calculated at user equipments (UEs). Second, we consider a different approach where the MSE can be calculated at base stations (BSs) i.e., leakage-based MSE from each BS. Different from conventional per-base station power constraint (PBPC), we assume practical per-antenna power constraint (PAPC) due to linearity of the power amplifier that feeds each transmit antenna at the BSs. To this end, we derive robust precoders and receive filters for these two design objectives under statistical channel state information (CSI) uncertainty. Simulation results suggest that our robust designs offer a good performance, showing resilience against CSI uncertainty. In addition, the sum-rate performances for designs with PAPC and PBPC are compared. It is observed that the sum-rate loss due to PAPC is relatively small for the robust designs.