Bandwidth Minimization under Probabilistic Constraints and Statistical CSI for NOMA

Abstract

Non-Orthogonal Multiple Access (NOMA) is studied under statistical Channel State Information (CSI) and probabilistic constraints. Unlike the conventional power domain only NOMA, the definition of NOMA here considers both the power and the frequency domains simultaneously, where a flat power spectrum is assumed for each UE. This increases the capacity region when compared to the conventional definition. The problem of minimizing the number of consumed resource elements (REs) while satisfying the outage constraints is solved. An RE here is a frequency slot of unit bandwidth resolution, so the considered problem translates to a saving in bandwidth. The solution techniques involve a Gaussian Approximation of NOMA user rates and an offline empirical approach based on tail distributions. Numerical results show that the former is a good approximation while the latter provides exact results. In any case, these methods outperform the orthogonal multiple access techniques.