Fuzzy Learning based Robust Channel Allocation in V2X Networks

Vehicle-to-everything (V2X) communication has become a core technology in intelligent transportation systems (ITS), providing essential connectivity between vehicles, infrastructure, pedestrians, and networks. However, the high mobility of vehicles introduces significant uncertainty in channel state information, making robust channel allocation a challenging problem. This paper proposes a robust V2X channel allocation scheme based on fuzzy learning. First, the dynamical and uncertain channel state information is characterized using fuzzy numbers. On this basis, the channel allocation process is carried out within a mapped fuzzy space. By integrating fuzzy logic with game based learning, the allocation task is formulated as a fuzzy payoff game (FPG), in which a fuzzy payoff function (FPF) is designed to represent the achievable rate under channel uncertainty. Furthermore, the fuzzy Nash equilibrium is derived to enable robust channel allocation. Building on this framework, the proposed robust fuzzy learning algorithm is employed to reach the equilibrium solution by leveraging the priority vectors obtained in the fuzzy space. Simulation results demonstrate that the proposed scheme significantly outperforms several benchmark methods, including the deep reinforcement learning-based DDPG algorithm, the crisp game-theoretic approach, and the random strategy. Specifically, the proposed method reduces the fluctuation range of the achievable rate by 90.8% and 91.4% compared to the crisp game and random strategy, respectively. In terms of the average achievable rate, the proposed scheme achieves performance improvements of 10.19%, 41.4%, and 99.6% over the three benchmark methods. Moreover, across different vehicle speeds and various clustering configurations, the proposed scheme consistently demonstrates superior performance.