Multi-cell MIMO user rate balancing with imperfect CSIT: SESIP vs. RESIP

In this work, we consider the max-min user rate balancing problem w.r.t. imperfect Channel Knowledge at the Transmitter (CSIT), namely: expected user rate balancing. This combines an operation of balancing at the user level and sum rate maximization at the level of the user streams. For the imperfect CSIT, we exploit an approximation of the expected rate as the Expected Signal and Interference Power (ESIP) rate, based on an original minorizer for every individual rate term. We study the latter with two expected rate approximations: i) Received signal level ESIP (RESIP), which may seem the most natural, and ii) Stream level ESIP (SESIP), which requires some more work for the stream level power optimization. Simulation results confirm the intuition that SESIP outperforms RESIP when the number of streams is lower than the number of receive antennas.