Covert wireless communication ensures both information confidentiality and transmission untraceability, which is increasingly vital for mission-critical extended reality (XR) services. While unmanned aerial vehicles (UAVs) provide mobility and flexible coverage, and intelligent reflecting surfaces (IRSs) enable energy-efficient signal manipulation, their joint use for covert communications has not yet been sufficiently explored. This paper proposes a novel UAV-mounted IRS system for covert communications that passively reflects source signals toward a legitimate receiver while minimizing detection by an adversary warden. In contrast to previous work that treats trajectory design, beamforming, and power control in isolation, the proposed work develops a unified framework based on double deep Q-networks (DDQN) to jointly optimize the UAV trajectory, power allocation, and IRS phase shifts under covert constraints. We analytically derive the optimal detection threshold and the minimum detection error probability, which are dynamically integrated into the learning framework. The optimization problem is formulated as a constrained Markov decision process, which allows the agent to adaptively learn optimal policies in dynamic environments without relying on perfect channel knowledge. Simulation results demonstrate that the proposed framework significantly improves covert rate and energy efficiency compared with the iterative and random benchmark schemes, while also providing insights into the impact of system parameters on performance.

The escalating demands for secure wireless communications in the Internet of Everything envisioned for the 6G era emphasize the urgency of advanced security solutions beyond traditional methods, especially in extended reality applications where secure wireless communications are essential for functionality and user experience. This paper explores the integration of covert communication techniques with deep reinforcement learning to bolster security in wireless networks. Covert communication, which prevents adversaries from detecting transmissions, is a critical factor in protecting data transmission over vulnerable wireless channels. This paper considers a two-hop wireless system model that is optimized using a double-deep Q-network algorithm. The problem is formulated as a constrained Markov decision process, jointly optimizing relay selection, transmission, and jamming powers to maximize covert communication rates while minimizing detection by adversarial wardens. Comprehensive numerical analysis demonstrates the effectiveness of the proposed method under various system conditions, including different configurations of relay and jamming powers. The results confirm that our model aligns well with theoretical expectations and substantially enhances covert communication by intelligently adapting to environmental dynamics.

Extended reality (XR) and 6G networks are set to transform mobile immersive experiences, with privacy and security being paramount in XR communications. Achieving secure and reliable XR experiences while meeting high-resolution and low-latency requirements is challenging for wireless networks. A novel security-aware cross-layer communication management framework is proposed, employing zero-trust spatiotemporal physical layer level manipulations for moving-target defense. Driven by deep reinforcement learning and real-time monitoring, the proposed framework adaptively reprograms the network configuration to maximize the user’s quality of experience (QoE), reduce the overall latency, and minimize the attacker’s intercept probability. The framework was evaluated in a simulated scenario featuring an indirect multi-hop communication setup. The results show that the proposed framework effectively and efficiently secures XR user communications while maintaining QoE, outperforming conventional Q-learning algorithms.

Considering the highly dynamic nature of the vehicular environment and the delay-sensitive wireless medium, enabling secure and reliable vehicle-to-vehicle communication becomes a very challenging task. In this paper, we propose a novel system design that uses blockchain technology to establish a high-level trust-management successful trustworthy cooperative vehicular wireless communication. The proposed system is a cross-layer approach that optimizes the best-relay selection process allowing only trustworthy relays to participate in data transmission. In this work, blockchain stores real-time information about relaying, transmitting, and receiving vehicles. The information includes channel characteristics and participation quality. The information is vetted and verified by mining vehicles. The result guides the relay selection process to block untrustworthy vehicles from participation. The entire system is comprehensively modeled and mathematically analyzed. Simulations using throughput and Bit Error Rate (BER) as evaluation metrics demonstrated the effectiveness and efficacy of the presented approach in enabling reliable wireless communication in presence of maliciously behaving vehicles. On average, the overall system throughput rate increased by$3bps$, the BER was reduced by almost by$2dB$, and the percentage of false messages decreased by 90% considering high Signal to Noise Ratio (SNR).

This paper proposes a novel approach to enhance wireless vehicle‐to‐vehicle channel‐secrecy capacity by imposing signal transmission diversity. This work exploits cooperative vehicular relaying to extract the associated underlying multipath and Doppler diversity using precoding techniques. We evaluated the capacity and diversity gain for the presented approach to ensure its effectiveness and efficiency. The abundance of moving vehicles, operating in an ad hoc fashion, can eliminate the need to establish a dedicated relaying infrastructure. A relay selection scheme is deployed, taking advantage of the potentially large number of available relaying vehicles. Further, we derivate a closed‐form mathematical expression for the channel‐secrecy capacity, diversity order gain, and the intercept probability. We used the direct transmission scenario as a reference to assess our analysis. Our analytical and simulation results for the presented model showed that channel‐secrecy capacity and performance‐indicators improved significantly.