In this paper, we study power allocation for downlink multi-carrier non-orthogonal multiple access (MC-NOMA) systems and examine the effects of residual cancellation errors resulting from imperfect successive interference cancellation (SIC) on the system performance. In the presence of random SIC errors, we study outage probability of minimum reserved rate for individual user and formulate outage-constrained robust optimization to minimize the total transmit power. Since the problem is non-convex due to probabilistic constraints, complementary geometric programming (CGP) and arithmetic geometric mean approximation (AGMA) technique are employed to transform it into a convex form. An efficient iterative algorithm with low computational complexity is developed to solve the optimization problem. Simulation results demonstrate the performance of robust MC-NOMA with imperfect SIC and compare that to non-robust MC-NOMA and orthogonal multiple access (OMA) schemes.

In this treatise, we derive tractable closed-form expressions for the outage probability of the single cell multicarrier non-orthogonal multiple access (MC-NOMA) downlink, where the transmitter side only has statistical CSI knowledge. In particular, we analyze the outage probability with respect to the total data rates (summed over all subcarriers), given a minimum target rate for the individual users. The calculation of outage probability for the distant user is challenging, since the total rate expression is given by the sum of logarithmic functions of the ratio between two shifted exponential random variables, which are dependent. In order to derive the closed-form outage probability expressions both for two subcarriers and for a general case of multiple subcarriers, efficient approximations are proposed. The probability density function (PDF) of the product of shifted exponential distributions can be determined for the near user by the Mellin transform and the generalized upper incomplete Fox's H function. Based on this PDF, the corresponding outage probability is presented. Finally, the accuracy of our outage analysis is verified by simulation results.

No abstract

No abstract