Driving Towards Efficiency: Adaptive Resource-Aware Clustered Federated Learning in Vehicular Networks

Guaranteeing precise perception for au-tonomous driving systems in diverse driving conditions requires continuous improvement and training of the perception models. In vehicular networks, federated learning (FL) facilitates this by enabling model training without sharing raw sensory data. Based on federated learning, clustered federated learning reduces communication overhead and aligns well with the dynamic nature of these networks. However, current literature on this topic does not consider critical aspects, including (1) the correlation between perception performance and the networking overhead, (2) the limited data storage on vehicles, (3) the need for training with freshly captured data, and (4) the impact of data heterogeneity (non-IID) and varying traffic densities. To fill these research gaps, we introduce AR-CFL, an Adaptive Resource-aware Clustered Federated Learning framework. AR-CFL dynam-ically enhances system efficiency by adaptively adjusting the number of clusters and specific in-cluster participant selection strategies. Using AR-CFL, we systematically study the online detection model training scenario on non-IID data across varied conditions. The evaluation results highlight the robust detection performance exhibited by the trained model employing the clustered federated learning approach, despite the constraints posed by limited vehicle storage capacity. Furthermore, our study reveals that utilizing clustered feder-ated learning enhances the training efficiency of participating nodes by up to 25% and decreases cellular communication by 33 % in contrast to conventional federated learning methods.