The evolution of sixth-generation (6G) networks enables transformative edge Artificial Intelligence of Things (AIoT) applications but introduces critical security vulnerabilities during model transmission between the central server and edge devices (e.g., unmanned aerial vehicles). Traditional approaches fail to jointly optimize model accuracy and physical layer security against eavesdropping. To address this gap, we propose a novel dynamic user selection framework that integrates three key innovations: (1) closed-form secrecy outage probability derivation for Rayleigh fading channels, (2) a Secure Model Accuracy (SMA) metric unifying recognition accuracy and secrecy outage probability, and (3) an alternating optimization algorithm for joint model–bandwidth selection under secrecy constraints. Comprehensive simulations demonstrate 22% SMA gains over baselines across diverse channel conditions and eavesdropper capabilities, resolving the fundamental accuracy–security tradeoff for trustworthy edge intelligence.

Hierarchical Federated Learning (HFL) alleviates the trade-off between communication overhead and privacy protection in mobile scenarios via multi-level aggregation and mobility consideration. However, its idealized convex loss assumption and full-dimension parameter upload deviate from real-world non-convex tasks and edge channel constraints, causing excessive energy consumption, high communication cost, and compromised convergence that hinder practical deployment. To address these issues in mobile/UAV networks, this paper proposes an energy-efficient optimization scheme for HFL under non-convex loss, integrating a dynamically adjustable partial-dimension model upload mechanism. By screening key update dimensions, the scheme reduces uploaded data volume. We construct a total energy minimization model that incorporates communication/computation energy formulas related to upload dimensions and introduces an attendance rate constraint to guarantee learning performance. Using Lyapunov optimization, the long-term optimization problem is transformed into single-round solvable subproblems, with a step-by-step strategy balancing minimal energy consumption and model accuracy. Simulation results show that compared with the original HFL algorithm, our proposed scheme achieves significant energy reduction while maintaining high test accuracy, verifying the positive impact of mobility on system performance.