Assessing the trustworthiness of network elements, including devices, communication links, and end-to-end paths, is critical for building secure and reliable networks. However, the heterogeneity of these components and the dynamic nature of their security states make trust assessment in large-scale networks highly challenging. This paper presents a configurable end-to-end network trustworthiness assessment scheme featuring a hierarchical, life-cycle-based architecture for evaluating trust from initial network deployment through long-term operation to retirement. In the proposed scheme, trustworthiness is assessed at three levels: device, link, and path. At the device level, a Dempster-Shafer evidence theory-based model integrates multi-dimensional trust attributes to quantify each device’s trustworthiness. At the link level, a dynamic multi-attribute model combines static and dynamic trust attributes and employs a clustering algorithm to calculate link trust scores. At the path level, a distributed fusion-based model uses fuzzy logic to evaluate end-to-end path trust by aggregating the trust levels of constituent devices and links. Simulation results demonstrate that the proposed scheme significantly improves both the accuracy and efficiency of trustworthiness assessment in large-scale networks.

Generative artificial intelligence (GAI), e.g., diffusion model is recognized as a promising paradigm for enhancing intelligent transportation systems in vehicular networks. However, the existing implementation of GAI in vehicular networks is limited due to the massive data requirements of GAI and the considerable resources for model training, particularly in distributed vehicular network environments. Federated learning (FL) offers a promising solution by enabling distributed collaborative training for GAI. Therefore, in this paper we present an FL-empowered diffusion model training scheme for vehicular networks. Specifically, first, a novel utility evaluation model based on local model training accuracy is designed to assess the contribution of each vehicle's local model. The interactions between the edge computing servers and vehicles are modeled using a Stackelberg game, while a non-cooperative game determines the optimal strategy among vehicles. To account for the heterogeneity of vehicles and the uncertainty of associated risks, we incorporate prospect theory (PT) to represent subjective utility. Afterward, a backward induction mechanism is devised to determine the Stackelberg equilibrium for deriving the optimal decisions of edge computing servers and vehicles. Finally, simulations are conducted to illustrate that the proposed scheme significantly improves the sum utility rate in comparison to other baseline schemes.

Recently, quantum federated learning (QFL) is advocated to leverage the robust computing power of quantum edge computing devices (QECDs) within unmanned aerial vehicle (UAV)-assisted wireless networks, to enhance the efficiency of distributed learning. However, the presence of malicious and selfish behaviors among some QECDs poses significant challenges for QFL model training to achieve high accuracy and rapid convergence. To tackle this issue, we introduce a trustenhanced incentive scheme for QFL in the UAV-assisted wireless networks. Specifically, a QECD-empowered QFL framework is first presented in the UAV-assisted wireless networks, where the QECDs independently train local models with their private data by using the quantum computing capabilities, while UAVs aggregate these trained local models to update the global model. Then, to ensure security and eliminate malicious participants, we devise a Bayesian inference-based trust assessment mechanism to select honest QECDs for local model training. Furthermore, we design a Stackelberg game-based incentive mechanism to incentivize QECDs to cooperatively provide high-quality training services. Afterwards, through game analysis using the backward induction method, we prove the existence of a Stackelberg equilibrium. The optimal payment strategies of the UAVs are obtained using the deep Q-learning network (DQN) algorithm in dynamic networks, and the optimal training contribution strategy of each QECD is derived using the convex optimization method. Finally, extensive simulations demonstrate that the proposed scheme can significantly enhance the accuracy and training speed of QFL in UAV-assisted wireless networks.

Aerial intelligent reflecting surface (AIRS), integrating unmanned aerial vehicle (UAV) with IRS, has emerged as a promising paradigm to improve the transmission quality and security in emergency communication, space-air-ground-integrated network and mobile edge computing, etc. However, the size of a single AIRS is constrained by the limited energy and payload capacity of the UAV, as well as the path loss of the air-to-ground reflective link, which makes the gain from a single AIRS finite. To address these problems, we propose a hybrid aerial IRS-assisted cooperative secure transmission system, where an aerial active IRS and an aerial simultaneously transmitting and reflecting IRS (STAR-IRS) are employed to achieve reflection amplification and 360-degree ubiquitous coverage, respectively. Additionally, the cooperative beamforming gain generated by the secondary reflection between the hybrid AIRSs can further improve communication quality. Specifically, an optimization problem is proposed with the objective of maximizing the sum secrecy rate by jointly optimizing the transmitting beamforming and the reflection coefficients of each AIRS. We first reformulated the original non-convex problem by fractional programming method, and a three-layer alternating optimization algorithm is introduced to address the proposed problem with the successive convex approximation (SCA) as well as penalty convex-concave procedure (PCCP) techniques. Finally, extensive simulations are conducted to demonstrate that the proposed scheme substantially improves the sum secrecy rate compared to other baseline schemes.