Fast Algorithm for Optimal Resource Allocation in a Low-Latency Downlink NOMA-Enabled Vehicle Platoon-Based V2X System

Vehicle-to-Everything (V2X) communication enables timely exchange of safety and control information among vehicles, roadside units, and the broader network, and is therefore central to realizing stable, high-density platooning under real-world conditions. In this work, we study low-latency downlink resource allocation for non-orthogonal multiple access (NOMA)-enabled vehicular platoons while accounting for realistic mobility and sensing imperfections. Unlike rate- or throughput-centric approaches, latency is treated as the primary objective, which is optimized subject to per-user quality of service (QoS) constraints as well as both global and per-platoon power budgets. The main contributions are as follows: (i) explicit incorporation of dynamic vehicle velocities and an imperfect-CSI model so that allocations adapt to mobility-induced channel fluctuations and estimated delays; (ii) a latency-minimization formulation that integrates the minimum-rate and power bounds per platoon together with the dual-level power constraints to reflect the practical infrastructure and platoon limitations; (iii) an analytical convex reformulation that provides provable optimality and enables predictable real-time operation; and (iv) closed-form intra-platoon power allocations derived from Karush–Kuhn–Tucker (KKT) conditions that involve the LambertWfunction, yielding interpretable per-vehicle rules and explicit minimum-platoon power expressions for admission control. Numerical results indicate that the proposed NOMA scheme yields dramatic latency improvements compared with other recent benchmarks, while the individual NOMA rates comfortably exceed the minimum threshold. Moreover, the average latency grows only modestly with increasing numbers of users and platoons, and raising the minimum-rate target is shown to further reduce latency. The proposed bisection-based algorithm converges rapidly (error typically below 10−6in about 10 iterations) and cuts down floating-point operations (FLOPs) by a factor on the order of the number of users and number of platoons when compared with other recent methods.