Online Adaptive Transmission Strategy for Buffer-Aided Cooperative NOMA Systems

This paper investigates a buffer-aided cooperative NOMA system, where a source is connected to multiple destinations through a buffer-aided NOMA relaying. We aim to maximize a long-term average network utility by jointly optimizing working mode selection, admission control, and power allocation in each time slot, subject to queue stability constraints and both the peak and average power constraints at the relay. To solve this problem, by leveraging the recently-developed Lyapunov optimization framework, we design an Online Adaptive Transmission (OAT) strategy, which converts the original long-term optimization problem into a series of online admission control and power allocation subproblems in each time slot, without loss of optimality. By considering the fact that the power allocation problem is non-convex in nature and thus cannot be solved with a standard method, we transform it into an equivalent convex optimization based on the specific feature of the formulation, and then propose a Fast Decision Criteria (FDC) based algorithm to derive the optimal solution with high computational efficiency. Extensive simulation results are provided to evaluate the performance of the proposed OAT strategy with FDC based power allocation for the buffer-aided cooperative NOMA system, and demonstrate that the proposed strategy can significantly outperform multiple benchmark schemes.