Outage-constrained resource allocation in uplink NOMA for critical applications

In this work, we consider the resource allocation problem for uplink non-orthogonal multiple access (NOMA) networks whose users represent power-restricted but high priority devices, such as those used in sensor networks supporting health and public safety applications. Such systems require high reliability and robust resource allocation techniques are needed to ensure performance. We examine the impact on system and user performance due to residual cancellation errors resulting from imperfect successive interference cancellation (SIC) and apply the chance-constrained robust optimization approach to tackle this type of error. In particular, we derive an expression for the user outage probability as a function of SIC error variance. This result is used to formulate a robust joint resource allocation problem that minimizes user transmit power subject to rate and outage constraints of critical applications. As the proposed optimization problem is inherently non-convex and NP-hard, we apply the techniques of variable relaxation and complementary geometric programming to develop a computationally tractable two-step iterative algorithm based on successive convex approximation. Simulation results demonstrate that, even for high levels of SIC error, the proposed robust algorithm for NOMA outperforms the traditional orthogonal multiple access case in terms of user transmit power and overall system density, i.e., serving more users over fewer sub-carriers. The chance-constrained approach necessitates a power-robustness trade-off compared to non-robust NOMA but effectively enforces maximum user outage and can result in transmit power savings when users can accept a higher probability of outage.