Multi-Agent Reinforcement Learning for C-V2X RAT Selection

Radio resource allocation (RRA) is a critical function in cellular vehicle-to-everything (C-V2X) networks, where vehicles must share limited wireless resources to support safety-critical communications. Multi-agent reinforcement learning (MARL) has emerged as a promising approach for this problem. However, key MARL challenges, including non-stationarity, coordination difficulty, large action space, partial observability, and limited robustness and generalization, are often intertwined, making it difficult to assess their individual impact on performance in vehicular environments. Moreover, existing studies primarily focus on developing new algorithms, while systematic benchmarking and comparative analyses remain limited. To address this gap, we formulate C-V2X RRA as a hierarchy of multi-agent interference games that progressively introduce key MARL challenges. Based on this framework, we develop a suite of benchmark learning tasks and construct training and testing datasets from SUMO-generated highway traces with diverse vehicular topologies and interference conditions. Using the proposed benchmark, we evaluate representative MARL algorithms spanning value-based, actor-critic, Independent Learning (IL), and Centralized Training with Decentralized Execution (CTDE) paradigms. The results identify robustness and generalization across diverse vehicular topologies as the dominant challenge among those considered in this work, reducing average normalized return by up to 59 percentage points, and show that, on the most challenging task, the best actor-critic method outperforms the best value-based method by 42%. By revealing the relative strengths and limitations of different MARL paradigms and open-sourcing the code, datasets, and benchmark suite, this work provides a systematic and reproducible foundation for evaluating and advancing MARL algorithms in vehicular networks.

Reliable, low-latency uplink connectivity is a key requirement for C-V2X networks in dense urban environments, where fast channel variations and blockages often degrade direct vehicle-to-infrastructure links. Multi-hop relaying can restore coverage, but relay-link activation under radio, capacity, and routing constraints results in an NP-hard optimisation problem, typically solved via Mixed-Integer Linear Programming (MILP), whose runtime scales poorly with graph size. This paper introduces an edge-aware Learning-to-Optimise framework for real-time relay selection. Each V2X snapshot is modelled as a directed graph: node features encode vehicle state and traffic demand, while edge features capture radio-link capacity. An offline MILP oracle generates optimal relay configurations that supervise a Graph Isomorphism Network with Edge Features (GINE), enabling edge-level relay activation through a single forward pass, with tightly bounded inference latency. To bridge learning and exact optimisation, we also propose a hybrid GINE-Pruned MILP (GP-MILP) strategy in which GINE predictions prune the MILP search space. Experiments on a large-scale dataset generated via an OSM-SUMO-GEMV$^2$ pipeline show that GINE closely matches MILP decisions at the link level (accuracy 0.9589), F1-score (0.9544) on validation) and yields consistent end-to-end connectivity gains over a 1-hop MILP baseline (up to 9.2% with four RSUs and 12% with two RSUs). Inference latency remains tightly bounded, with all evaluated instances completing within 5ms. Moreover, GP-MILP preserves MILP-equivalent solutions (same objective value) while achieving solver runtimes below 30ms for more than 98%) of the graph instances, making MILP-grade optimisation compatible with stringent NR-V2X latency budgets.

The 3GPP V2X resource allocation framework defines two entity classes -- the base station and the vehicle UE -- and four modes across LTE and NR generations. We demonstrate that this binary taxonomy is structurally incomplete. Base station-led scheduling saturates at high-density traffic nodes, producing latency-tail failures that persist even when mean packet delivery ratios approach the service-class target. UE autonomy is categorically incapable of pre-emergence warning for occluded traffic participants and insufficient for large-scope cascading environmental hazards. We propose Mode 0, a new 3GPP V2X category whose defining entity is the Roadside Computing Unit (RCU) -- an infrastructure ensemble integrating elevated sensing (Seeing), sidelink communication (Speaking), and local computational evaluation (Thinking), owned by traffic management authorities. Mode 0 defines a subfamily spectrum from Mode 0a (all-passive UEs, the guaranteed minimum) through Mode 0c (all-active UEs, the optimal target). Convergent deployment evidence from Chinese national standards (DB11/T 2329.1-2024, T/ITS 0224.1-2025), China Unicom RS-MEC infrastructure, and European and US C-V2X programs confirms that both institutional sides are converging on the roadside traffic node without a coordination standard. A fifteen-run Multi-Agent Proximal Policy Optimization (MAPPO) simulation validates the architectural family: Mode 0a in shared-pool baseline sits at the analytical symmetric-Nash coordination floor; Mode 0c with demand separation achieves strict Pareto improvement for both traffic classes (M0 PDR 0.999, M1 PDR 0.998 at $ρ_{\rm pool} \leq 1$) and lifts the worst-TTI delivery ratio from near-zero to 0.601 -- the only configuration satisfying the latency safety requirement structurally. We call for a 3GPP study item on Mode 0 within the NR-V2X sidelink enhancement work programme.