With the increasing demand for data security in high-speed railway (HSR) communications, physical layer security (PLS) has emerged as an effective approach to enhance the secrecy performance of wireless systems. This paper investigates a millimeter-wave secure communication system for HSR scenarios, assisted by intelligent reflecting surface (IRS), where the channel state information (CSI) of the eavesdropper (Eve) is unavailable. The primary objective is to minimize the secrecy outage probability (SOP). First, a closed-form expression of SOP with imperfect CSI is derived based on a statistical channel model, which provides a theoretical foundation for performance evaluation. Then, to address the difficulty of channel prediction in high-mobility scenarios, an attention-enhanced long short-term memory (Attention-LSTM) model is proposed to predict the CSI of the mobile relay (MR), effectively mitigating the channel aging issue caused by CSI latency. To tackle the issues of non-convex optimization and tightly coupled variables in the joint design of transmit beamforming and IRS phase adjustments, a twin delayed deep deterministic policy gradient (TD3) algorithm is employed to derive an effective secrecy enhancement strategy for HSR communications. Simulation outcomes indicate that the proposed framework achieves superior secrecy performance compared to conventional optimization methods, highlighting its robustness and practical applicability.

High-speed railway (HSR) communication is subject to unauthorized eavesdropping, and the acquisition of perfect channel state information (CSI) is difficult, which increases the secrecy outage probability of the system and poses a threat to secure data transmission. In addition, the train’s operating environment is complex. Deviations in train speed can increase Doppler shift compensation errors, further increases the secrecy outage probability. This paper constructs a train speed prediction model based on the L-Ns-Transformer. By compensating for Doppler shift, the model achieves more accurate channel modeling. Furthermore, an intelligent reflecting surface (IRS)-assisted beamforming method for HSR secure communication is proposed when the eavesdropper’s (Eve) CSI is unknown. We propose a two-stage deep learning (TS-DL)-based approach to design transmitter beamforming and IRS jointly, where the precoding vector and phase shift matrix are designed to minimize the secrecy outage probability. Simulation results demonstrate that the proposed TS-DL approach has lower computational complexity and can effectively reduce the system secrecy outage probability, thereby enhancing the security of HSR wireless communication.

As the railway industry undergoes digital transformation driven by AI, big data, and 5G/6G, secure data transmission for IoT-oriented railway operations has become a critical challenge in high-speed railway (HSR) wireless communications. Covert communication, which conceals wireless transmissions from eavesdroppers, together with reconfigurable intelligent surfaces (RISs), MIMO, and millimeter-wave technologies, offers a promising solution for enhancing security in such dynamic environments. This paper presents a comprehensive review of RIS-assisted covert communication for mmWave MIMO channels in HSR IoT networks. We examine security threats and the limitations of existing schemes, introduce the system architecture and channel modeling approaches, and summarize Doppler shift compensation techniques for high-mobility scenarios. In particular, this review introduces a service-priority-based covert constraint design framework as an original analytical perspective, rather than a fully validated engineering solution, to improve adaptability in diverse HSR operational contexts. We further provide a critical comparative review of optimization algorithms, including convex relaxation, heuristic, and deep reinforcement learning approaches, with a focus on their applicability and complexity trade-offs, and outline future research directions. Beyond summarizing the existing literature, this review integrates new analytical perspectives and simulation-based illustrations derived from our recent publications, offering both a systematic reference and forward-looking insights for researchers in this emerging field.

In multi-scenario Mobile Edge Computing (MEC), offloading and resource allocation for topologically heterogeneous Directed Acyclic Graph (DAG) tasks are NP-hard, while Federated Meta-Learning (FML) suffers from generalization degradation under non-independent and identically distributed data. To address both challenges, this paper proposes a Difficulty-Aware FML (DAFML) algorithm. We construct a three-dimensional difficulty metric—integrating task load, topological complexity, and resource supply–demand tension—and determine its weights via the Analytic Hierarchy Process. Building on this metric, a difficulty-guided non-uniform meta-task sampling strategy prioritizes complex tasks, while Proximal Policy Optimization (PPO) handles inner-loop adaptation. The offloading problem is formulated as minimizing a normalized composite cost of latency, energy, and deadline violations. Simulation results show that DAFML achieves a composite cost of 0.8096 and a completion rate of 82.61%, outperforming Greedy, centralized PPO, FedAvg-PPO, and Fed-MAML. Compared with Fed-MAML, DAFML reduces the composite cost by 2.43% and improves the completion rate by 2.26 percentage points, with greater gains on complex subsets. Ablation studies validate the contributions of the sampling strategy and the three difficulty dimensions, and convergence analysis reveals that DAFML has the most stable trajectory among all algorithms.