Xiao-Lei Wang

Accurate and robust indoor localization is essential for advanced wireless communication, navigation, and mobility services, yet multipath propagation and dynamic channel variations pose significant challenges. This paper proposes STFNet, a deep neural architecture designed for joint time-frequency modeling of channel state information (CSI). STFNet adopts a dual-branch architecture, where a temporal module captures hierarchical motion dynamics and a spectral attention module extracts salient structural features from the CSI spectrum. A gated fusion mechanism is further introduced to adaptively integrate temporal and spectral representations, thereby enhancing accuracy and robustness. In addition, an adaptive multi-task loss is formulated to jointly optimize localization accuracy and trajectory smoothness, while a weighted moving average is employed to mitigate short-term inference noise. Extensive experiments across diverse indoor scenarios demonstrate that STFNet achieves sub-meter localization accuracy, with mean absolute error ranging from 0.45 m to 0.55 m. Furthermore, STFNet achieves superior performance compared with representative and state-of-the-art methods in terms of accuracy, stability, and generalization, indicating strong potential for real-world deployment.

Accurate indoor localization is critical for emerging Internet of Things and smart environment applications. Channel State Information (CSI), with its fine spatial resolution and high sensitivity to environmental changes, offers strong potential in Wi-Fi-based indoor localization. However, practical deployment is hindered by multipath interference and device heterogeneity, which degrade positioning accuracy and system robustness. To address these challenges, this paper proposes a deep learning-enhanced CSI acquisition and processing framework, termed MSRANet. The framework incorporates parallel multi-scale temporal convolutions to capture hierarchical and fine-grained features, deep residual blocks for enhanced feature representation, and a dual attention mechanism to refine frequency and spatial features. A lightweight Weighted Moving Average module is also introduced to smooth prediction outputs and improve stability. The proposed system is implemented on low-cost embedded hardware and evaluated in multiple real-world indoor environments. Experimental results show a mean absolute error of around 0.7 meters, demonstrating improved positioning accuracy and robustness compared to existing methods, while maintaining computational efficiency suitable for real-time deployment.

This study aims to address the challenge of backfill compaction in the confined spaces of municipal utility tunnel trenches and to develop an environmentally friendly, zero-cement-based backfill material. The research focuses on the excavation slag soil from a utility tunnel project in Handan. An alkali-activated industrial-solid-waste-excavated slag-soil-based controllable low-strength material (CLSM) was developed, using NaOH as the activator, a slag–fly ash composite system as the binder, and steel slag-excavated slag as the fine aggregate. The effects of the water-to-solid ratio (0.40–0.45) and the binder-to-sand ratio (0.20–0.40) on CLSM fluidity were studied to determine optimal values for these parameters. Additionally, the influence of excavated soil content (45–65%), slag content (30–70%), and NaOH content (1–5%) on fluidity (flowability and bleeding rate) and mechanical properties (3-day, 7-day, and 28-day unconfined compressive strength (UCS)) was investigated. The results showed that when the water-to-solid ratio is 0.445 and the binder-to-sand ratio is 0.30, the material meets both experimental and practical requirements. CLSM fluidity was mainly influenced by the excavated soil and slag contents, while NaOH content had minimal effect. The unconfined compressive strength at different curing ages was negatively correlated with the excavated soil content, while it was positively correlated with slag and NaOH content. Based on these findings, the preparation of “zero-cement” CLSM using industrial solid waste and excavation slag is feasible. For trench backfill projects, a mix of 50–60% excavated soil, 40–60% slag, and 3–5% NaOH is recommended for optimal engineering performance. CLSM is a new type of green backfill material that uses excavated soil and industrial solid waste to prepare alkali-activated materials. It can effectively increase the amount of excavated soil and alleviate energy consumption. This is conducive to the reuse of resources, environmental protection, and sustainable development.

In this article, the problem of piecewise fault detection (FD) for continuous-time piecewise linear (PWL) system with measurement outliers via new event-triggered (ET) mechanisms is investigated. First, piecewise FD observer with adaptive saturation of output errors is designed to mitigate the effect of outliers on the residual signals, such that false alarms are avoided through the established FD scheme. Then, based on the presented piecewise FD observer, a new ET design scheme that depends on the different regions where the states of the PWL system and piecewise FD observer are located is proposed, and an ET FD method is studied through the ET mechanisms. Unlike the existing results, the established FD scheme can effectively avoid false alarm caused by outliers, and the piecewise FD observer is designed separately from the new ET mechanisms, while co-design is necessary in the existing results. Finally, the effectiveness of the presented scheme is demonstrated by simulation examples.

This article studies the design of fault detection (FD) observers for nonlinear steering control (NSC) systems of unmanned surface vehicles (USVs) with steering machine faults. The considered NSC systems of USVs are characterized by discrete‐time T‐S fuzzy models with immeasurable states. First, a new fuzzy FD observer is constructed to handle the challenge caused by the introduction of membership functions (MFs) that are different from the ones in the system, in which the MFs in the FD are designed by scaling and biasing those of the T‐S model. Different from the existing uncertainty restrictions between the MFs in the systems and those in the observers, the deterministic restrictions are constructed, and the MFs of the observer are directly constructed according to the ones of the system without calculation. Then, by employing Lyapunov function (LF), it is proved that the constructed FD observers can ensure the stability and desired performance of the systems. Finally, an NSC system model of USVs is given to verify the effectiveness of the presented technical solution.