Energy-Efficient Massive MIMO Multi-Relay NOMA Systems With CSI Errors

We consider a massive multi-input multi-output (mMIMO) system where in a base station (BS) serves its users via multiple relays by employing non-orthogonal multiple access (NOMA). We practically model this system by considering channel estimation errors both at the BS and at the users, which consequently perform imperfect successive interference cancellation. We consider these two artifacts, and derive a lower bound on the spectral efficiency (SE) of this multi-relay NOMA system, which is valid for arbitrary number of BS antennas. We then jointly allocate the BS and relay powers to optimize non-convex global energy efficiency (GEE) and the weighted sum energy efficiency (WSEE) metrics, which are fractional functions of sum-of-products/ratios of optimization variables. These two optimizations require extension of an existing fractional programming framework, which we do by proposing two novel transformations. We analytically prove the monotonic convergence of the transformed problems to a stationary point of their respective original counterparts. We numerically show that i) a multi-relay mMIMO NOMA system outperforms its orthogonal counterpart only with accurate channel information; and ii) the proposed GEE and WSEE algorithms significantly improve the energy efficiency by avoiding additional power use, after attaining optimal value.