SINR Distributions for Robust MIMO TX-Power Minimization With Controlled Outage

Abstract

We offer a simple and comprehensive alternative to existing robust network optimization methods as a solution to the long-standing problem of cellular network optimization under imperfect channel state information at the transmitter (CSIT) conditions. Specifically, instead of relying on a relaxation of statistical constraints into deterministic convex equivalents, as done in current literature and which leads to tradeoffs between robustness and optimality, we derive accurate models for the distributions of users’ signal-to-interference-plus-noise ratio (SINRs), conditioned on constraint-matching SINR estimates obtained under imperfect CSIT. Using these distributions, precise accounting and compensation for imperfection on CSIT, as well as statistical outage matching, can be achieved directly via the design of deterministic, instantaneous, and convex SINR constraints. Simulated comparison with the state of the art demonstrates the significant performance gain achieved by the proposed solution.