Convenient indoor positioning has become an urgent need due to the improvement it offers to quality of life, which inspires researchers to focus on device-free indoor location. In areas covered with Wi-Fi, people in different locations will to varying degrees have an impact on the transmission of channel state information (CSI) of Wi-Fi signals. Because space is divided into several small regions, the idea of classification is used to locate. Therefore, a novel localization algorithm is put forward in this paper based on Deep Neural Networks (DNN) and a multi-model integration strategy. The approach consists of three stages. First, the local outlier factor (LOF), the anomaly detection algorithm, is used to correct the abnormal data. Second, in the training phase, 3 DNN models are trained to classify the region fingerprints by taking advantage of the processed CSI data from 3 antennas. Third, in the testing phase, a model fusion method named group method of data handling (GMDH) is adopted to integrate 3 predicted results of multiple models and give the final position result. The test-bed experiment was conducted in an empty corridor, and final positioning accuracy reached at least 97%.

Wireless sensor networks (WSNs) have been used in many fields due to its wide applicability. In this kind of network, each node is independent of each other and has its own local clock and communicates wirelessly. Time synchronization plays a vital role in WSNs and it can ensure accuracy requirements for coordination and data reliability. However, two key challenges exist in large-scale WSNs that are severe resource constraints overhead and multihop time synchronization errors. To address these issues, this paper proposes a novel unmanned aerial vehicle (UAV)-assisted low-consumption time synchronization algorithm based on cross-technology communication (CTC) for a large-scale WSN. This algorithm uses a UAV to send time synchronization data packets for calibration. Moreover, to ensure coverage and a high success rate for UAV data transmission, we use CTC for time synchronization. Without any relays, a high-power time synchronization packet can be sent by a UAV to achieve the time synchronization of low-power sensors. This algorithm can achieve accurate time synchronization with almost zero energy consumption for the sensor nodes. Finally, we implemented our algorithm with 30 low-power RF-CC2430 ZigBee nodes and a Da Jiang Innovations (DJI) M100 UAV on a 1 km highway and an indoor site. The results show that time synchronization can be achieved accurately with almost zero energy consumption for the sensor nodes, and the time synchronization error is less than 30 $\mathsf{\mu }$s in 99% of cases.

Ultra-dense wireless cellular networks have been envisioned as a promising technique for handling the explosive increase of wireless traffic volume. With the extensive deployment of small cells in wireless cellular networks, the network spectral efficiency (SE) is improved with the use of limited frequency. However, the mutual inter-tier and intra-tier interference between or among small cells and macro cells becomes serious. On the other hand, more chances for potential cooperation among different cells are introduced. Energy efficiency (EE) has become one of the most important problems for future wireless networks. This paper proposes a cooperative bargaining game-based method for comprehensive EE management in an ultra-dense wireless cellular network, which highlights the complicated interference influence on energy-saving challenges and the power-coordination process among small cells and macro cells. Especially, a unified EE utility with the consideration of the interference mitigation is proposed to jointly address the SE, the deployment efficiency (DE), and the EE. In particular, closed-form power-coordination solutions for the optimal EE are derived to show the convergence property of the algorithm. Moreover, a simplified algorithm is presented to reduce the complexity of the signaling overhead, which is significant for ultra-dense small cells. Finally, numerical simulations are provided to illustrate the efficiency of the proposed cooperative bargaining game-based and simplified schemes.

Long-term multivariate time series forecasting is crucial for real-world applications, including energy consumption, traffic flow, healthcare, and finance. Usually, some statistical approaches are used for predicting future observations based on historical temporal data. Recently, transformer-based models with patch mechanisms have demonstrated significant potential in enhancing computational efficiency. However, their inability to fully capture intra-patch temporal dependencies often limits the accuracy of predictions. To address this issue, we propose the Frequency Compensation Patch-wise transFormer (FCP-Former), which integrates a frequency compensation layer into the patching mechanism. This layer extracts frequency-domain features via Fast Fourier Transform (FFT) and incorporates them into patched data, thereby enriching patch representations and mitigating intra-patch information loss. To verify the feasibility of this model, FCP-Former was conducted on eight benchmark datasets via PyTorch 2.4.0 and trained on an NVIDIA RTX 4090 GPU (Santa Clara, CA, USA). Results demonstrate that FCP-Former 48 optimal experiment results and 17 suboptimal experiment results across all datasets. Especially on the ETTh1 dataset, it achieves an average MSE of 0.437 and an average MAE of 0.430, while on the Electricity dataset, it achieves an average MSE of 0.186 and an average MAE of 0.277. This demonstrates that FCP-Former has better forecasting accuracy and a superior ability to capture periodic and trend patterns.

Time series forecasting can be applied to various aspects of daily life, providing valuable insights to inform decision-making in areas such as stock market, electricity load, traffic flow and health-care, among others. Many previous studies have demonstrated the effectiveness of transformer-based models in long-term time series forecasting tasks, and the introduction of patch mechanisms can further enhance the predictive performance of Transformer-based models. However, most of the previous works have focused on reducing the computational resource overhead of patch-wise models, with little consideration given to the potential decrease in the model's ability to capture information from each time step within the patches. This often limits the further improvement of prediction accuracy in patch-wise models. To Address this issue, we propose a frequency compensation block, which incorporates frequency-domain data features into each patch before performing the time series forecasting task, to improve patch-wise model, called Frequency Compensation Patch-wise transFormer(FCP-Former). Experimental results demonstrate that the proposed method achieve better performance compared with the state-of-the-art methods in multivariate time series forecasting tasks.

The demand for robust safety monitoring in underground coal mines is increasing, yet traditional methods face limitations in long-term stability due to inadequate energy supply and high maintenance requirements. To address the critical challenges of high computational demand and energy constraints in this resource-limited environment, this paper proposes MineVisual, a battery-free visual sensing scheme specifically designed for underground coal mines. The core of MineVisual is an optimized lightweight deep neural network employing depthwise separable convolution modules to enhance computational efficiency and reduce energy consumption. Crucially, we introduce an energy-aware dynamic pruning network (EADP-Net) ensuring a sustained inference accuracy and energy efficiency across fluctuating power conditions. The system integrates supercapacitor buffering and voltage regulation for stable operation under wind intermittency. Experimental validation demonstrates that MineVisual achieves high accuracy (e.g., 91.5% Top-1 on mine-specific tasks under high power) while significantly enhancing the energy efficiency (reducing inference energy to 6.89 mJ under low power) and robustness under varying wind speeds. This work provides an effective technical pathway for intelligent safety monitoring in complex underground environments and conclusively proves the feasibility of battery-free deep learning inference in extreme settings like coal mines.