Due to the large-scale, dynamic topology and multi-hop communication characteristics of Internet of Things (IoT) systems, trust management has become a critical requirement for secure and reliable network operation. However, existing trust management schemes generally rely on endpoint-centric observations or centralized evidence aggregation, making them poorly suited to realistic IoT deployments with partial observability. In particular, performance degradations caused by compromised intermediate nodes or unstable links are often misattributed to benign devices, leading to distorted trust evaluation and inefficient trust-based control. To address these challenges, this study proposes a topology-aware trust management framework (GZTrust) for multi-hop IoT networks designed in accordance with Zero Trust principles. The GZTrust framework introduces a structured separation between trust evidence generation, trust reasoning, and trust-driven control. In the data plane, a TrustTrace mechanism is employed to incrementally collect hop-level trust evidence along forwarding paths with verifiable integrity and privacy-preserving pseudonyms. In the trust plane, dynamically selected policy enforcement points (PEPs) aggregate localized evidence in a topology-aware manner to avoid flat centralized processing. To accurately evaluate trust, this study designs an attribution-aware trust computation model that decomposes hop-level anomalies into node-centric and link-centric responsibility components. Extensive simulation results demonstrate that GZTrust consistently outperforms state-of-the-art trust management schemes, improving the F1-score by approximately 6–13% across varying malicious node ratios (5%–50%). These results indicate that GZTrust provides a practical and scalable trust management framework for dynamic multi-hop IoT environments, with a lightweight overhead profile that is well aligned with resource-constrained deployment settings.

Path-aware networks have garnered significant attention as an emerging research area. It allows network senders to actively select or influence transmission paths to meet specific requirements, which necessitates the support of path validation mechanisms. Supported by the path-aware networking research group under the Internet Engineering Task Force (IETF), path validation plays a crucial role in enhancing end hosts' control over packet forwarding. However, existing methods face trade-offs among security, protocol header overhead, and computational cost, forming a ''trilemma.'' Drawing inspiration from persistent validation in zero-trust architecture, we propose the 1BIT protocol. This protocol reduces protocol header overhead by more than 57% while providing robust data flow security. The packet demand for path fault detection is reduced by more than 72%, and fault locations can be precisely identified. By employing hash algorithms and few binary operations, the 1BIT protocol achieves high throughput and supports routers capable of adapting to high-speed, multi-interface environments. On a 16-core CPU, the 1BIT protocol can handle throughput exceeding 100 Gbps. This lightweight and efficient solution introduces anomaly signal detection techniques into the field of path validation. Benefiting from in-depth research on anomaly signal detection, this technology offers a richer set of solutions for path validation and lays the foundation for future research and implementation in areas such as multi-path validation and path privacy protection.

Due to characteristics such as the openness and exposure of inter-satellite links, Low-Earth Orbit (LEO) satellite networks are subject to heightened vulnerability to malicious attacks compared to ground-based networks. Given various security risks, implementing trust management in LEO satellite networks becomes imperative. However, existing trust management schemes tailored for this scenario are vulnerable to a spectrum of attacks, resulting in low detection rates and poor network performance. To address the above issues, we propose BiTrust, a hybrid trust management scheme for secure data transmission in LEO satellite networks. BiTrust introduces two distinct types of trust: state trust and behavior trust. State trust leverages remote attestation technology to verify the authenticity of node identities and the integrity of their functions, ensuring that nodes consistently disseminate reliable behavior trust announcements to safeguard the trust plane. Behavior trust, on the other hand, utilizes a trust model to quantify the real-time data forwarding behavior characteristics of nodes, thereby maintaining the reliability of the data plane. To precisely quantify behavior trust, we design a trust model based on multi-path trust propagation. By integrating an unstable penalty term, the proposed trust model can effectively defend against dynamic dropping misbehavior. Besides, to facilitate the deployment of BiTrust in a distributed manner, we introduce a two-hop rely message mechanism and a query-answer mechanism to support the construction of behavior trust. Experimental results confirm the efficacy of BiTrust, demonstrating a significant improvement of up to 64.3% in data transmission rate under highly untrusted environments while maintaining affordable overhead.

Popular semi-supervised image segmentation networks suf-fer from two problems: firstly, supervision is only performed on the last layer of the decoder, resulting in the network's weak generalization ability; secondly, the geometry shape constraints of targets are frequently disregarded in these net-works, leading to poor segmentation results. To address these issues, we propose a novel shape-guided dual consistency semi-supervised learning framework for 3D medical image segmentation. The proposed framework makes two contri-butions. Initially, we introduce a shape constraint to learn the shape representation, which converts the difference be-tween two networks into an unsupervised loss and lets the model learn the boundary information of targets. Addition-ally, we develop a deep-supervised knowledge transfer strat-egy that improves the generalization ability of the network without increasing extra computation costs. Experiments demonstrate that the proposed framework outperforms state-of-the-art semi-supervised methods due to the strong ability of knowledge mining on unlabeled data.

The popular Siamese convolutional neural networks (CNNs) for remote sensing (RS) image change detection (CD) often suffer from two problems. First, they either ignore the original information of bitemporal images or insufficiently utilize the difference information between bitemporal images, which leads to the low tightness of the changed objects. Second, Siamese CNNs always employ dual-branch encoders for CD, which increases computational cost. To address the above issues, this article proposes a network based on difference enhancement and spatial–spectral nonlocal (DESSN) for CD in very-high-resolution (VHR) images. This article makes threefold contributions. First, we design a difference enhancement (DE) module that can effectively learn the difference representation between foreground and background to reduce the impact of irrelevant changes on the detection results. Second, we present a spatial–spectral nonlocal (SSN) module that is different from vanilla nonlocal because multiscale spatial global features are incorporated to model the large-scale variation of objects during CD. The module can be used to strengthen the edge integrity and internal tightness of changed objects. Third, the asymmetric double convolution with Ghost (ADCG) module is exploited instead of standard convolution. The ADCG can not only refine the edge information of the changed objects, since horizontal and vertical convolutional kernels have good contour preservation advantages, but also greatly reduce the computational complexity of the proposed model. The experiments on two public VHR CD datasets demonstrate that the proposed network can provide higher detection accuracy and requires smaller memory usage than state-of-the-art networks.