Hui Zhang

This paper investigates a simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS) assisted backscatter communication system. Our objective is to minimize the transmit power by jointly optimizing the transmit beamforming vector, the tags' reflection coefficients, the STAR-RIS coefficient matrices and energy splitting coefficients at the STAR-RIS, subject to the signal-to-interference-plus-noise ratio constraints for decoding backscatter signals and the power constraints of the tags. We propose an alternating optimization algorithm to find near-optimal solution. Numerical results show that our proposed scheme can significantly reduce the transmit power even under the assumption that all reflection/transmission elements have the same amplitude coefficients.

In this paper, we investigate an unmanned aerial vehicle (UAV) assisted integrated sensing and communication (ISAC) system, where reconfigurable intelligent surface (RIS) is mounted on and powered by the UAV. We aim to maximize the average achievable rate via jointly optimizing the UAV trajectory and beamforming at both the base station (BS) and the RIS with varying environments. While deep reinforcement learning (DRL) is capable of solving stochastic optimizations, it lacks generalization in dynamic context. The decision Transformer (DT) is then designed to solve the generalization issue of DRL. However, it shows poor performance in capturing the short range local dependency patterns that arise in trajectories governed by a Markov decision process (MDP). To address these limitations, we propose a hybrid decision causal Convformer (HDCC). We first formulate the optimization problem with the framework of MDP. Then we utilize two complementary parallel branches in our HDCC, i.e. self-attention for modeling long range global dependencies and convolution for extracting short range local temporal patterns. Next, we propose a causal block embedded in the convolution branch. This is to identify intrinsic causal relations among convolution features and convert them into an auxiliary action proposal, improving action prediction and policy robustness. Simulation results show that the proposed HDCC achieves an improvement of 9.4% in terms of average transmission rate, compared to the conventional DT.

Wi-Fi-based human activity recognition (HAR) has emerged as a focal point within the Internet of Things landscape, due to its nonintrusive sensing capabilities and inherent privacy-preserving advantages. In existing Wi-Fi-based HAR research, channel state information (CSI) is primarily utilized to capture activity-related features and enable recognition. However, CSI-based cross-domain HAR remains challenged by issues, such as redundant subcarriers, limited samples in the target domain, and high sensitivity of CSI to environmental variations. To address these challenges, this article proposes Wi-DMAR, a Wi-Fi-based cross-domain HAR framework that integrates three key modules. First, an adaptive subcarrier selection module computes the correlation between each subcarrier and the principal components, identifies subcarriers with high contribution, preserves essential activity-related features, reduces data dimensionality, and lowers computational overhead. Second, a conditional diffusion-based data augmentation module employs a Transformer-based feature extractor to capture domain-specific representations of the target-domain data, and optimizes the domain consistency loss and domain-guided diffusion loss to generate pseudo-samples that resemble the target-domain distribution, thereby mitigating the sample scarcity. Third, an activity recognition module based on sample similarity learning reformulates the traditional label classification problem into a sample comparison task, by quantifying similarity between samples, and it performs the activity recognition and enhances the cross-domain generalization. Experimental results demonstrate that Wi-DMAR achieves the superior recognition accuracy compared with state-of-the-art cross-domain HAR methods, such as DiffAR and MetaAct. Ablation studies further confirm that each core component contributes positively to performance improvements.

This study explores an industrial Internet of Things (IIoT) scenario characterized by the dense deployment of femtocells within a macrocell coverage area. To address the critical challenges of inter-device interference and spectral efficiency optimization, we propose a novel D2D group architecture that integrates non-orthogonal multiple access (NOMA) technology for efficient group communications. However, the complex cumulative interference caused by channel sharing in such a network severely impairs the system performance, which was seldom considered in the previous study. To mitigate this issue, our framework proposes an innovative hypergraph-based spectrum allocation algorithm, where femtocell access points (FAPs) are represented as vertices and interference relationships are captured as hyperedges, enabling the application of hypergraph theory to formulate the optimal resource allocation problem. Comprehensive experimental evaluations demonstrate that the proposed scheme significantly improves the spectral efficiency compared to traditional graph-based spectrum allocation schemes.

To enhance the efficiency and safety of mobile robot path planning, this letter proposes a hybrid sampling and space-optimized RRT (HB-RRT) method. First, a hybrid sampling strategy combining Gaussian sampling with parallel sampling is employed to replace conventional random sampling. During the sampling process, the sampling space is dynamically optimized by adjusting the sampling region to the improve efficiency. Then, a potential field-guided expansion strategy with adaptive step size is introduced to reduce excessive exploration and limit expansion in collision-prone areas by leveraging environmental and node information. The expansion exchange strategy is refined to improve stability, and a direct connection method is adopted to rapidly generate an initial path. Furthermore, a path optimization method is designed to refine the initial path. Finally, a series of experiments are conducted to validate the proposed method against baseline and state-of-the-art algorithms. The results show that the proposed method improves planning efficiency and reduces path cost in generating smooth paths.

Digital twin (DT) and mobile edge computing (MEC) are two promising solutions for achieving latency-sensitive and computing-intensive applications in Industrial Internet of Things (IIoT). However, existing collaborative task offloading schemes with DT empowerment are faced with challenges, such as the complex collaborative relationship between tasks and multiple edge servers (ESs), and spatio-temporal heterogeneity of ES resources. This article investigates the issues of parallel collaborative offloading and resource allocation under the assistance of the DT and overlapping coalition formation (OCF) game. One novel scheme, abbreviated as OCF-based PCORA, is proposed. With comprehensive information within the digital space, the OCF-based PCORA scheme dynamically models collaborative relationships between tasks and ESs. Subsequently, each task is offloaded to its optimal ES coalition for efficient collaborative processing. The offloading request is formulated as a nonconvex problem. To make the nonconvex problem solvable in polynomial time, the original problem is decomposed into three subproblems: ED association, bandwidth allocation, and collaborative task processing. The first subproblem is transformed through slack variable relaxation and solved using the interior-point method. The second subproblem, being naturally convex, is addressed via a convex optimization method. The third subproblem is modeled as an OCF game with transferable utility. On top of this, a bilevel iterative optimization algorithm is proposed to form overlapping task coalitions in a distributed manner. Numerical results demonstrate that the proposed scheme reduces the average task completion latency by 16.01%–44.97% and decreases the task offloading failure rate by 28.99%–75.41%, compared to state-of-the-art baselines.

In this paper, we propose an innovative spectrum sharing incentive mechanism tailored for Unmanned Aerial Vehicle (UAV) swarms, utilizing a Stackelberg game framework. This mechanism enables multiple UAV-to-UAV (U2U) links to effectively share the spectrum resources of a single UAV-to-Base Station (U2B) link. We commence by designing a hybrid overlay-underlay spectrum access mode, addressing the challenge of inter-user interference. Subsequently, the spectrum sharing issue is formulated as a Stackelberg game model, where U2B links are positioned as leaders and U2U links as followers. This model facilitates an interactive environment where both entities converge on an optimal trading strategy, encompassing the determination of the ideal trading price and allocation of resource blocks (RBs). To further enhance this system, we employ a utility-based matching strategy. This strategy ensures that U2U links select the U2B link that optimizes their utility, while the U2B link, in turn, chooses the U2U link that maximizes its utility from the pool of applicants seeking spectrum sharing. Simulation results underscore the effectiveness of this approach, showcasing significant improvements in both system utility and the per-unit utility of spectrum resources. Additionally, our mechanism successfully incentivizes more UAV users to engage in a mutually beneficial spectrum sharing arrangement.

The path planning is one of the critical technologies for robots to achieve the autonomous operation, enabling to quickly find a safe path in dynamic environments. However, relying on global path planning alone cannot avoid dynamic obstacles, while only using the local path planning may lead to falling into local minima. Therefore, a two-layer robot path planning method suitable for dynamic environments is proposed. This two-layer strategy consists of an efficient global path planning layer and a safe local dynamic obstacle avoidance layer. In the first layer, a parallel sampling and bidirectional guidance rapidly exploring random tree algorithm (PB-RRT) is proposed to search for the global path. To enhance efficiency, the parallel heuristic sampling is introduced to replace the random sampling in bidirectional rapidly exploring random tree (Bi-RRT), and an evaluation function incorporating distance and corner factors is designed to select optimal sampling nodes for adaptive expansion, making the sampling process directional and avoiding over-exploration of space. A bidirectional guidance mechanism further accelerates the merging of the two trees by fully utilizing newly generated node information. Then, a path optimization (PO) method is proposed to improve the length and smoothness of the initial path and obtain the key nodes of the path. In the second layer, the key nodes obtained from the first layer are used as dynamic subtargets, and the safe dynamic window approach (SDWA) is used to achieve the dynamic obstacle avoidance. To further enhance safety, an offset guidance method is proposed to flexibly steer the robot around dynamic obstacles. Extensive experiments show that the average planning time of PB-rapidly exploring random tree (RRT) is reduced by 67.7% compared with Bi-RRT, while the path quality is improved and the path length is reduced by 27.1%. The proposed method also effectively avoids dynamic obstacles in environments with obstacle densities exceeding 60% and achieves the maximum of minimum distance to dynamic obstacles, validating the feasibility and safety of the method.