A User Pairing Algorithm for the Uplink NOMA With Imperfect CSI

ABSTRACT In this paper, the problem of user pairing across subcarriers is investigated for the uplink non‐orthogonal multiple‐access (NOMA) systems, in which every subcarrier is assigned to exactly two users. Considering fixed transmit power for all users, we pair users such that sum‐rate is maximized over all subcarriers. While different (near) optimum allocation schemes exist in the literature, they pertain to perfect channel state information (CSI), which is impractical. To overcome this limitation, we assume statistical CSI is available in the form of the first two moments of channel gains. Then, we maximize a lower bound on the sum rate over user pairings utilizing a block coordinate ascent (BCA) approach with guaranteed convergence to a suboptimal solution. Numerical results corroborate a satisfactory performance for our proposed algorithm versus the perfect CSI benchmark and a simple random assignment approach. For high signal to noise ratios (SNRs) and the more challenging frequency selective fading setup, the solution of our proposed algorithm comes close to those of the optimal pairing with perfect CSI. Furthermore, simulations revealed that if the proposed BCA is utilized assuming perfect CSI, it performs similar to the well‐known global optimum in the frequency flat fading setup and outperforms the best existing near‐optimal solution in the frequency selective fading scenario. These observations offer a testament to the strength of the proposed BCA algorithm in both perfect and imperfect CSI conditions.