EE Optimization for Downlink NOMA-Based Multi-Tier CRANs

Non-orthogonal multiple access (NOMA) is increasingly becoming very attractive in cloud radio access networks (CRANs) to further boost the overall spectral efficiency, connectivity and, capacity of such networks. This paper addresses optimizing the energy efficiency (EE) for the downlink of a NOMA-based two tiers CRAN. The stochastic geometry represented by Poisson Point Process (PPP) distribution is used to decide the number and locations of the base stations (BSs) in each tier within the coverage area. A numerical optimal solution is obtained and compared against a proposed subgradient solution, as well as another proposed unoptimized solution based on the false positioning method. For comparison purposes, two other power allocation techniques are presented to allocate different powers to various BS categories; one allocates the power to each BS based on their relative distances to the cloud-based central station and the other is the bisection based scheme. Two simulation scenarios are presented to examine the performance of the two-tier NOMA-CRANs with NOMA is adopted as the multiple access of each tier in both cases. The first scenario considers heterogeneous CRAN (NOMA-HCRAN) case by using two different BS categories in each tier, namely, the macro-BSs and the RRHs. The second scenario considers a homogeneous CRAN (NOMA-CRAN) case by using the RRHs in both tiers but each tier has different frequency layer to prevent cross tier interference. Simulation results show the promising performance gain can be achieved with the proposed techniques relative to the existing approaches. More specifically, it was illustrated that the proposed subgradient based NOMA CRAN offers better performance than the proposed false positioning based NOMA CRAN, which is in turn better than the existing techniques, in particular, the bisection and the distance based NOMA-CRAN.