The deployment of Federated Learning (FL) in edge environments is often impeded by system heterogeneity, non-independent and identically distributed (non-IID) data, and constrained communication resources, which collectively hinder training efficiency and scalability. To address these challenges, this paper presents FLEX-SFL, a flexible and efficient split federated learning framework that jointly optimizes model partitioning, client selection, and communication scheduling. FLEX-SFL incorporates three coordinated mechanisms: a device-aware adaptive segmentation strategy that dynamically adjusts model partition points based on client computational capacity to mitigate straggler effects; an entropy-driven client selection algorithm that promotes data representativeness by leveraging label distribution entropy; and a hierarchical local asynchronous aggregation scheme that enables asynchronous intra-cluster and inter-cluster model updates to improve training throughput and reduce communication latency. We theoretically establish the convergence properties of FLEX-SFL under convex settings and analyze the influence of local update frequency and client participation on convergence bounds. Extensive experiments on benchmark datasets including FMNIST, CIFAR-10, and CIFAR-100 demonstrate that FLEX-SFL consistently outperforms state-of-the-art FL and split FL baselines in terms of model accuracy, convergence speed, and resource efficiency, particularly under high degrees of statistical and system heterogeneity. These results validate the effectiveness and practicality of FLEX-SFL for real-world edge intelligent systems.
Resource constraints, time-varying links, and heterogeneous data distributions make distributed model training difficult in unmanned aerial vehicle (UAV) networks. Conventional federated learning can impose substantial computational and communication burdens on resource-constrained UAVs, whereas existing split-learning schemes commonly rely on fixed split points and static participation strategies. To address these limitations, this paper proposes HCFSL-UAV, a hierarchical cluster-head federated split learning framework for UAV networks. The framework adopts a three-layer architecture composed of a central base station, cluster-head UAVs, and member UAVs, and integrates split learning within hierarchical federated aggregation. Member UAVs are dynamically associated with cluster heads according to their own real-time 3-D spatial locations. Within each cluster, participant selection and model splitting are jointly optimized as the set of selected UAVs affects data representativeness and synchronous training latency, while the split point determines the computation, memory, and communication requirements of participating UAVs. A joint optimization problem is formulated to balance data-distribution deviation, training latency, and resource consumption, and an alternating optimization procedure is developed using greedy participant selection and a discrete split-point search. Experiments on FMNIST, CIFAR-10, and CIFAR-100 show that HCFSL-UAV achieves faster convergence and higher test accuracy than the considered federated-learning and split-learning baselines, while reducing communication and computation overhead in the reported settings. These results demonstrate the effectiveness of jointly optimizing participant selection and adaptive model partitioning in dynamic and heterogeneous UAV networks.