Paper detail

Probability-constrained Power Optimization for Multiuser MISO Systems with Imperfect CSI: A Bernstein Approximation Approach

We consider power allocations in downlink cellular wireless systems where the basestations are equipped with multiple transmit antennas and the mobile users are equipped with single receive antennas. Such systems can be modeled as multiuser MISO systems. We assume that the multi-antenna transmitters employ some fixed beamformers to transmit data, and the objective is to optimize the power allocation for different users to satisfy certain QoS constraints, with imperfect transmitter-side channel state information (CSI). Specifically, for MISO interference channels, we consider the transmit power minimization problem and the max-min SINR problem. For MISO broadcast channels, we consider the MSE-constrained transmit power minimization problem. All these problems are formulated as probability-constrained optimization problems. We make use of the Bernstein approximation to conservatively transform the probabilistic constraints into deterministic ones, and consequently convert the original stochastic optimization problems into convex optimization problems. However, the transformed problems cannot be straightforwardly solved using standard solver, since one of the constraints is itself an optimization problem. We employ the long-step logarithmic barrier cutting plane (LLBCP) algorithm to overcome difficulty. Extensive simulation results are provided to demonstrate the effectiveness of the proposed method, and the performance advantage over some existing methods.