Dynamic Power and Rate Allocation for NOMA Based Vehicle-to-Infrastructure Communications

In this paper, a non-orthogonal multiple access (NOMA) based downlink vehicle-to-infrastructure network is considered. Particularly, we focus on the specific case of two users, one of which requires reliable road-safety-critical data transmission while the other pursues high-capacity services, with extension to multi-user scenarios. Leveraging only slow fading of channel state information, the transmit powers and target rates are jointly optimized to maximize the expected sum throughput of the capacity hungry user, with consideration of the payload delivery outage probability of the reliability sensitive user. The optimization is formulated as an unconstrained single-objective sequential decision problem via introducing a dual variable. A dynamic programming based algorithm is then designed to derive the optimal policy that maximizes the Lagrangian. Afterwards, a bisection search based method is proposed to find the optimal dual variable. The proposed scheme is shown by numerical results to be superior to the baseline methods in terms of the expected return, performance region, and objective value.