Region of Interest (ROI) segmentation is a core technology for image semantic understanding and visual attention analysis, widely applied in fields such as image retrieval, advertising design, and autonomous driving perception. Traditional methods rely on low-level visual features of images and ignore the subjective orientation of human visual attention, resulting in deviations between segmentation results and actual attention regions. To address this issue, this paper proposes a fixation duration-weighted ROI segmentation system for natural images. By collecting eye-tracking data to extract key parameters, constructing heatmaps through fixation point clustering and duration weighting, and integrating semantic features with adaptive threshold segmentation, high-precision segmentation is achieved, providing a new solution for intelligent ROI segmentation. Furthermore, to break through the dependence constraint of the original system on eye-tracking data, this paper proposes a subsequent optimization scheme: using the fixation duration-weighted heatmaps of the Toronto dataset as the supervisory ground truth to train a deep network that is independent of eye-tracking data. This scheme enables eye-tracking hardware-free ROI segmentation, greatly enhancing the deployment value and application scope of the research results.

This paper investigates the extended Kalman filtering issue for a class of time-varying nonlinear systems equipped with energy harvesting sensors under a modified stochastic communication protocol (SCP). Considering the energy consumption in the data transmission, the energy harvesting sensors are deployed which are able to capture renewable energy from the external environment to sustain data transmission. Moreover, to cater for the energy variation of the energy harvesting sensors, a modified SCP is proposed in which the transition probability matrix is dynamically adjusted according to the sensors’ energy levels and historical transmission records. Our aim is to develop an extended Kalman filtering algorithm for the underlying systems such that a minimal upper bound (UB) is ensured on the filtering error covariance (FEC). First, an UB on the FEC is obtained. Then, the filter gain that minimizes the obtained UB is designed. Finally, a simulation is provided to confirm the effectiveness of the proposed filtering algorithm.

The rapid development of the Internet of Things (IoT) and green communication systems has heightened the importance of backscatter communication (BackCom) and UAV-assisted communication. However, in UAV-assisted energy harvesting (EH)-based BackCom networks, system performance is significantly challenged by practical issues such as uncertain reflection coefficients (RCs) at backscatter sensors and imperfect channel state information (CSI). These uncertainties degrade both data transmission efficiency and communication reliability, thereby limiting the potential benefits of these advanced technologies. Motivated by the need to overcome these challenges, this work proposes a robust resource allocation framework that maximizes system throughput while satisfying throughput constraints. Specifically, our objective is to jointly optimize key parameters including BackCom time allocation, transmission power, primary user interference protection, and RC settings. To mitigate the adverse effects of uncertainties in CSI parameters and RCs, we introduce a robust learning algorithm based on polytope to transform interruption probability constraints into deterministic constraints. Finally, the resulting non-convex optimization problem is decomposed into two subproblems using an alternating optimization method, and subsequently solved via the Lagrange multiplier method. Simulation results show that the proposed algorithm achieves good network throughput and energy efficiency (EE) in UAV-assisted EH-based BackCom network.

In densely deployed massive Internet of Things (DD-mIoT) environments, device-to-device (D2D) users often transmit signals simultaneously, leading to significant signal interference and conflicts. These issues severely degrade communication quality and system performance, especially under limited communication resources. To address this challenge, we propose a conflict-free resource allocation scheme leveraging a Multi-Agent Deep Double Q-Network (MADDQN). First, we define interference weights for D2D users and model the interference using a hypergraph theory. Then, we introduce the concept of interference degree, derived from the hypergraph model, and reformulate the interference avoidance problem as the problem of satisfying the constrained interference degree. This problem is then formulated as a Markov Decision Process (MDP). Finally, we propose a resource allocation strategy based on the MADDQN algorithm to mitigate interference in DD-mIoT systems and enhance network performance. Simulation results demonstrate that the proposed scheme significantly improves network throughput and energy efficiency.

This paper proposes an attention-based neural network for OFDM channel estimation, designed to refine conventional least-squares (LS) estimates by exploiting time-frequency correlation characteristics of wireless channels. The proposed network incorporates a lightweight attention mechanism with query, key, and value mappings, enabling adaptive symbol-wise filtering while maintaining low computational complexity suitable for online deployment. An online training framework is introduced to adapt the network to time-varying channels with limited training samples. Simulation results demonstrate that the proposed method significantly outperforms conventional LS and MMSE estimators in terms of mean square error (MSE), especially under low signal-to-noise ratio (SNR) conditions. Furthermore, over-the-air experiments using a USRP testbed validate the practical feasibility and robustness of the method in realistic multipath fading environments.

Uncrewed aerial vehicle (UAV)-assisted intelligent transportation systems (ITS) are vital for smart cities but face significant challenges in energy efficiency (EE) and data timeliness. While existing studies have concentrated on maximizing EE, the critical issue of data timeliness has received less attention. This paper proposes an EE and age of information (AoI)-aware resource allocation (EA-RA) scheme. First, to quantify and manage data timeliness for diverse services, we formulate a precise AoI model by deriving the average AoI expression. Based on this model, we construct a stochastic optimization problem aimed at maximizing the network's EE, subject to the transmission rate requirements of various AoI-sensitive services and interference constraints. Finally, to handle the practical issue of imperfect channel state information (CSI) arising from quantization errors and feedback delays, the proposed EA-RA scheme solves the optimization problem by integrating Dinkelbach's method with Gaussian theory. Simulation results demonstrate that the proposed EA-RA algorithm not only satisfies the AoI requirements of different services but also concurrently improves the network's energy efficiency and data throughput.

With the emergence of information overload in the Social Internet of Things (SIoT), personalized recommender systems (RSs) have become essential for helping users locate the items they need. Effectively modeling the heterogeneous relationships across multiple information sources and weighting them according to their varying importance, especially under sparse social interactions in the SIoT, remains a key challenge. To address this issue, this article proposes a multisource adaptive relational graph convolutional network (MARGCN) framework for RSs. User-user and user-item interactions are modeled as two relational graphs, and their embedded features are aggregated using graph convolutional networks (GCNs). A multisource relationship perception model is designed to dynamically perceive and measure the importance of multiple relationships within the graph structure, thereby enhancing the ability to recognize heterogeneous information. An adaptive information fusion model is then constructed to dynamically integrate representations from different sources through a learnable fusion strategy, avoiding information loss or redundancy caused by simple weighting. Users and items are ultimately represented by aggregating and updating their embeddings via GCNs. Experiments show that compared with the most advanced methods, MARGCN improves the hit ratio (HR) and the normalized discounted cumulative gain (NDCG) by 2.66%, 0.52%, and 3.93% in HR@5, HR@10, and HR@15, and by 1.58%, 1.32%, and 3.14% in NDCG@5, NDCG@10, and NDCG@15, respectively.

In the smart city, high-density deployment of consumer electronics (CE) may lead to mutual interference, resulting in imperfect estimation of the channel state information (CSI). To tackle the problem, this paper proposes a split learning-based robust resource allocation for CEs in smart cities. We constructed an interference hypergraph model and divided resource allocation conflicts in overlapping areas into multiple virtual sub-cells (VSCs) to reduce the impact of mutual interference for the CSI. Then, we take into account the imperfect CSI and design a robust optimization model to maximize the throughput of the network in the VSCs. Due to the imperfections of CSI and the introduction of random channel parameters, solving robust optimization models is challenging. Hence, we propose the split robust learning algorithm based on interference hypergraph (SRLA-IH), which utilizes split learning theory to learn models and obtain more accurate uncertainty sets, effectively reducing the problems caused by imperfect CSI in smart cities. Numerical results demonstrate that compared with other algorithms, our proposed algorithm can achieve excellent network throughput and improve resource allocation utilization even under imperfect CSI.

This paper proposes an innovative event-triggered interval type-2 (IT-2) fuzzy consensus control method for agent-based full-vehicle suspension systems (FSSs) using a leader-following approach. By developing a novel IT-2 Takagi-Sugeno (T-S) fuzzy model, the amalgamation of heterogeneous agent-based FSSs into a cohesive homogeneous multi-agent framework becomes feasible, thereby reducing control complexity and effectively dealing with system uncertainty. Within the proposed agent-based architecture, a virtual leader is designed at the core of the agent-based FSS, and the four interconnected quarter-vehicle suspension systems are construed as the following agents. To optimize network bandwidth between the agents, a new event-triggered mechanism (ETM) is established, which is sensitive to significant state changes, particularly when deviations from consensus among the following agents arise abruptly. Sufficient conditions are derived to ensure both the optimal attitude performance and ride comfort of FSSs. Finally, a simulation example of agent-based FSSs is presented to validate the advantages of the proposed approach in optimizing ride comfort and system robustness.

With the development of the future Web of Healthcare Things (WoHT), there will be a trend of densely deploying medical sensors with massive simultaneous online communication requirements. The dense deployment and simultaneous online communication of massive medical sensors will inevitably generate overlapping interference. This will be extremely challenging to support data transmission at the medical-grade quality of service level. To handle the challenge, this paper proposes a hypergraph interference coordination-aided resource allocation based on the Deep Reinforcement Learning (DRL) method. Specifically, we build a novel hypergraph interference model for the considered WoHT by analyzing the impact of the overlapping interference. Due to the high complexity of directly solving the hypergraph interference model, the original resource allocation problem is converted into a sequential decision-making problem through the Markov Decision Process (MDP) modeling method. Then, a policy and value-based resource allocation algorithm is proposed to solve this problem under simultaneous online communication and dense deployment. In addition, to enhance the exploration ability of the optimal allocation strategy for the agent, we propose a resource allocation algorithm with an asynchronous parallel architecture. Simulation results verify that the proposed algorithms can achieve higher network throughput than the existing algorithms in the considered WoHT scenario.

This paper aims to discusses the distributed filtering problem under the gossip communication protocol under the consideration of fading phenomenon. In order to efficiently transmit, merge local data and save bandwidth, a gossip communication protocol is introduced, drawing inspiration from the social phenomenon of gossiping. In addition, the Rayleigh fading measurement is taken into account in the channel between the sensor and the filter. By applying the principles of Lyapunov stability theory, a set of matrix inequalities is established to guarantee the system’s boundedness in terms of mean square sense. A numerical example is given to demonstrate the effectiveness of the proposed results.

This paper proposes a theoretical model for a few-mode free-space optical communication (FSOC) reception based on optical-domain coherent beam combining. By analyzing the coupling efficiency of few-mode fiber, the relationship between atmospheric turbulence and coherent beam combining is established. The bit error rate (BER) performance is quantitatively evaluated by incorporating device loss factors and combining efficiency, thereby constructing a comprehensive theoretical framework that encompasses turbulent disturbance, mode coupling, and optical-domain beam combining across the entire signal transmission chain. Simulation and experimental results demonstrate that at a BER of 3.8×10 −3 , the proposed scheme achieves a 1–3 dB improvement in receiver sensitivity compared to single-mode FSOC systems, while performing within 1 dB of systems utilizing mode diversity reception technology. Moreover, the scheme requires only a single local oscillator and a single high-speed detector, significantly reducing the hardware complexity and noise accumulation associated with multi-branch detection.

Multiple Input Multiple Output (MIMO) technology is widely applied in various wireless communication systems, significantly improving communication efficiency and reliability. Signal detection is critical for MIMO systems. However, with the increasing integration of deep learning into MIMO signal detection algorithms, challenges such as high complexity and limited interpretability have emerged. To address this, this paper proposes a model driven trainable approximate message passing (AMP) algorithm that combines the iterative process of AMP with deep learning techniques. By introducing trainable parameters and optimizing them through training, and incorporating an attention mechanism to enhance channel feature extraction, the detection accuracy is improved, and the algorithm’s generalization capability is enhanced. Simulation results demonstrate that AMP Attention Net achieves lower bit error rates compared to traditional detection algorithms. Furthermore, the proposed algorithm exhibits robust performance under different configurations of transmitting and receiving antennas.

In the dynamic Low-Power Internet of Things (IoT) scenario with vehicle-to-vehicle (V2V) communications, each vehicle with TinyML will cause interference to others under a dynamic environment, resulting in cumulative overlapping interference and imperfect channel state information (CSI). Aiming at the problem of interference avoidance and resource allocation considering cumulative overlapping interference in large-scale V2V vehicular networks with imperfect CSI, this paper proposed an interference hypergraph-based resource allocation for large-scale Low-Power IoT with imperfect CSI. To analyze the complex cumulative overlapping interference, we construct the interference hypergraph model and obtain multiple virtual sub-cells. Then, a queue-based delay model is built for delay-sensitive services and a stochastic resource allocation model considering the quality of service (QoS) and imperfect CSI are established under the multiple virtual sub-cells. Since the introduced imperfect CSI with random parameters will make the model uncertain and hard to solve, we employ Gaussian theory to transform the uncertain constraints into deterministic constraints and propose a resource allocation scheme for large-scale Low-Power IoT with imperfect CSI. Simulation results demonstrate that the proposed scheme achieves high throughput and energy efficiency

In massive Internet of Vehicles (mIoV) networks, the substantial number of vehicles and the dense deployment of communication devices lead to extensive overlap in signal coverage, resulting in frequent co-channel interference (CCI) and adjacent-channel interference (ACI). This interference significantly impacts the Quality of Service (QoS) experienced by vehicles. Moreover, the QoS requirements for vehicles vary widely across different application scenarios, further complicating the design and optimization of resource allocation strategies. To this end, we propose a robust resource allocation method for differentiated QoS in mIoV, leveraging ellipsoid learning and joint interference management. First, a weighted interference hypergraph model is developed to effectively mitigate ACI and CCI experienced by vehicles, while simultaneously addressing the diverse QoS requirements across different vehicle applications. Then, a robust optimization model for vehicle-to-vehicle pair (VP) communications is proposed, and underlying spectrum sharing is adopted to improve spectrum utilization. Furthermore, an ellipsoid-based learning robust algorithm (ELRA) strategy is proposed to address dynamic uncertainties arising from imperfect channel state information (CSI), thereby enhancing the reliability and transmission rate of the communication link. Simulation results demonstrate that the proposed algorithm achieves superior network throughput and spectral efficiency performance in densely deployed mIoV.

In the Internet of Vehicles (IoV) with dense interference, overlapping coverage between vehicle user equipments’ communication radius (VUEs) leads to significant interference, causing resource reuse conflicts between transmission links. To address this, we construct an interference hypergraph model to analyze the interference relationships among multiple users through hyperedges. The degree of interference for the overall network is calculated using this model. The optimization problem for resource allocation is then formulated to minimize the data rate so that multiple quality of service (QoS) requirements are met at the same time while considering interference degree. Furthermore, a novel conflict graph model is constructed to represent the conflicts among transmission links. This model is then transformed into a conflict hypergraph structure, enabling a more comprehensive and accurate depiction of complex interference relationships within the network. We convert the combinatorial optimization problem for radio resource allocation into a Markov decision process (MDP) model. For IoV with dense interference, we employ a federated averaging-deep Q-Network (FedAvg-DQN) algorithm to enhance the transmission rate and overall network throughput. Consequently, the experimental findings demonstrate that this method can significantly enhance the anti-interference capacity of the system in a densely interfered Internet of Vehicles (IoV) environment. Specifically, compared to the benchmark algorithm, network throughput is on average increased by 28.17%, and energy efficiency is on average improved by 24.85%. These improvements effectively accelerate the information transmission rate and guarantee the reliability of the information transmission. These findings validate the efficacy and practicality of the developed approach, demonstrating its potential to enhance the performance of IoV systems under challenging interference conditions.

A novel and effective method for the synthesis of chiral C1-phosphinoylated 1H-isochromenes has been developed. The reaction proceeds via a tandem asymmetric addition/cyclization under a catalytic system of Pd(OAc)2, a Josiphos-based ligand L9, and ZnCl2 in MeCN at 70 °C. This method demonstrates a broad substrate scope (27 examples), delivering excellent yields (up to 92%) and enantioselectivities (up to 95%).

Microseismic event identification is the primary task of microseismic monitoring in hydraulic fracturing. Due to U-Net network will produce high time and space complexity in the training process, the learning efficiency of the model would be affected. Furthermore, the traditional cross entropy loss function is prone to interference from noisy samples when disposing microseismic data with a low signal-to-noise ratio (SNR), leading to a decrease in identification accuracy. In order to solve these problems, this paper proposes a microseismic event identification method using a lightweight U-Net network based on optimized cross entropy loss function. In this method, the label smoothing technology is used to optimize the cross entropy loss function, and the real label is adjusted by the smoothing coefficient, so that the model can better distinguish the effective signal from the noise features, and effectively suppress the oversensitivity of the cross entropy loss function to the noise features. The convolution layer in U-Net network is replaced by depthwise separable convolution (DSC), then the time and space complexity of the model can be reduced effectively through the synergistic effect of depthwise convolution and pointwise convolution, and the network model can be lightweight. After DSC, squeeze-and-excitation attention mechanism module is introduced to make the model focus on the channels with greater weight, which can improve the extraction ability and identification accuracy of effective features of the model. The experimental results show that, compared with the traditional U-Net network, the proposed model reduces the time and the space complexity by 64.6% and 65.2%. Also, the proposed model presents high identification accuracy and anti-noise ability when identifying true data with low SNR, and has good generalization ability.

In the distributed Cognitive Industrial Internet of Things (CIIoT), since industrial devices may self-organize to determine their connection and dispersion and the network may be distributed with no infrastructure, network management with self-organizing characteristics is still an unsolved and a difficult problem. To overcome this challenge, this paper proposes a hierarchical network management strategy for distributed CIIoT with imperfect channel state information (CSI) considering services with different transmission delay requirements. First, we build a two-layer architecture that supports collaborative spectrum sensing and multi-node spectrum sharing in a distributed CIIoT. Then, to improve the network management efficiency and reduce the number of backbone nodes (BNs) while achieving spectrum sharing for massively distributed nodes, we establish a minimum backbone node set mathematical model and design the simplest backbone network optimization algorithm (SBNOA). For the problem of spectrum sharing with imperfect CSI while considering the delay-sensitive industrial services, we establish a time delay model that incorporates delay-sensitive and delay-tolerant services. Based on the model, we establish a robust optimization model with multiple conditional aim-listed probability constraints that consider imperfect CSI and transform the stochastic optimization problem into a convex optimization by using the quadratic transformation method and Gaussian Q function to solve it. Simulation results show that the proposed algorithm has good performance in a distributed CIIoT.

Dear Editor, This letter focuses on the protocol-based non-fragile state estimation problem for a class of recurrent neural networks (RNNs). With the development of communication technology, the networked systems have received particular attentions. The networked system brings advantages such as easy to implement, high flexibility as well as low cost, and also has disadvantages such as limited bandwidth of the communication network which lead to networked-induced phenomena [1], [2]. To alleviate the network-induced phenomena, communication protocols have been introduced in the communication networks of the networked systems [3], [4]. As a widely used communication protocol in real practice, the round-robin (RR) protocol has received research interest and the state estimation problem under the RR protocol is an on-going hotspot in the area of signal processing [5]. Nevertheless, for the RNNs, the corresponding RR protocol-based state estimation problem still needs further research effort which is the first motivation of this letter.