Relay Power Allocation for NAF Cooperation Assisted NOMA Network

In this letter, a power allocation problem at secondary users (SUs) to maximize the overall system throughput is addressed in a non-orthogonal amplify-and-forward (NAF) cooperation assisted non-orthogonal multiple access (NOMA) network. Due to the nonlinearity of optimization problem, a cooperative multi-agent Q-Learning based power allocation algorithm (CQPA) is proposed under both sum and individual power constraints of SUs with relay function. To further reduce the computational complexity, a neighboring strategy searching based power allocation algorithm (NSPA) is additionally proposed. Simulation results demonstrate the efficiency of proposed algorithms, where a valuable trade-off between throughput and computational complexity can be achieved by the NSPA.