RSMA in Massive MIMO Systems: Analysis and Optimization in Correlated Rician-Faded Channels With Phase Shifts and Aging

We consider the downlink of a massive multi-input-multi-output (mMIMO) system where a base station (BS) serves ultra-reliable and low-latency communication (URLLC) user equipments (UEs) by employing rate-splitting multiple access (RSMA) technology. The UEs are assumed to be mobile, and their channels consequently age. The channels are spatially-correlated and Rician faded, with a phase-shifted line-of-sight (LoS) component, and the BS employs maximum ratio (MR) and minimum mean squared error (MMSE) precoders. A closed-form spectral efficiency (SE) expression is derived for the MR precoder, which is then used to develop a novel low-complexity optimal power allocation algorithm for the BS to maximize the global energy efficiency (GEE) metric. For the MMSE precoder, whose closed-form SE expression is difficult to derive, an iterative stochastic optimization algorithm is proposed. This algorithm, which is designed to have closed-form power update expressions, converges faster to the optimal solution than a conventional technique, which numerically calculates the SE. These low-complexity practically-implementable algorithms are developed by applying the epigraph, Lagrangian dual and quadratic transformations. We numerically show that RSMA outperforms spatial division multiple access when channels age faster, have a low Rician factor, and have a low correlation.