LWLiang Wang

Liang Wang

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Liang Wang has a total of 60 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Multi-Agent Reinforcement Learning for Joint Task Offloading and Migration in UAV Swarm Wireless Networks

    by Xiaolong Li, Fei Hao, Liang Wang, Liang Zhao

    The evolution of smart-city services demands sustainable and resilient network and service management in emergencies. In post-disaster scenarios where terrestrial infrastructure is damaged, mobile users face strict latency requirements under limited energy budgets and computing resources, while UAV-assisted edge networks experience dynamic topology and time-varying capacity. This paper studies the joint decision problem of task offloading, queue-aware scheduling, and inter-UAV task migration in UAV swarms under fixed communication and computation resource parameters, using a simplified distance-based communication model and UAV operational constraints. We propose an emergency edge computing framework and develop MAPPO-QM, a multi-agent reinforcement learning method based on Multi-Agent Proximal Policy Optimization that explicitly incorporates queueing dynamics and a rule-based migration mechanism triggered at migration points. MAPPO-QM optimizes a combined objective of end-to-end latency and energy consumption through an exponential reward with a fixed scaling coefficient, complemented by completion and cooperation reward terms. Simulation results show that MAPPO-QM achieves the highest task completion rate (99.13%) among the evaluated methods (PPO-QM, MAPPO, PPO, and local execution) and reduces the success-conditioned average latency by 58.2% compared with PPO-QM. MAPPO without the waiting queue and migration attains lower latency and energy per completed task but completes about 17% fewer tasks, indicating that queue-aware task waiting and migration can improve service availability in the evaluated post-disaster edge scenarios.

  • Open Access

    Robust Closed-Loop Control of Industrial Systems Based on Cloud–Edge Device Collaboration

    by Wenjing Zhang, Wenchao Zhang, Xiao Ma, Weijia Han, Liang Wang, Minghang Chen

    Under China’s dual-carbon strategic goals, large-scale public centralized heating plays a critical role in energy conservation. However, traditional manual open-loop control suffers from high response latency. Furthermore, existing unidirectional predictive methods lack dynamic feedback correction mechanisms. To address these issues, this study proposes an intelligent dual-system collaborative control architecture specifically designed for public heating systems. This architecture utilizes a cloud–edge device framework. It establishes a two-way linkage between heating equipment and the cloud decision platform. Consequently, it constructs an integrated regulation framework encompassing forward decision generation, reverse state verification, and dynamic feedback correction. Specifically, the forward module utilizes a Mamba-structured state-space model to generate data-driven boiler operation strategies. Meanwhile, the reverse module employs a Temporal Convolutional Network with Monte Carlo Dropout (TCN-MC Dropout). This probabilistic network enables state inversion evaluation with reliable uncertainty prediction intervals. These two modules are deeply coupled through an adaptive feedback correction mechanism. Together, they significantly improve system stability and operational robustness under complex thermal disturbances. Specifically, the proposed architecture achieves a room temperature compliance rate exceeding 96% and restricts temperature fluctuations to within ±0.75 °C. Simultaneously, it reduces boiler energy consumption by over 16.4%. This solution has been successfully deployed in the heating network at Shaanxi Normal University as a representative real-world case study. Ultimately, it provides a practical technical reference for the intelligent upgrading and low-carbon transformation of public centralized heating systems.

  • Multimodal Learning-Based Relational Graph Neural Networks for Social Bot Detection

    by Ke Wu, Yaguang Lin, Xiaoming Wang, Liang Wang

    Social bots are virtual accounts controlled by automated programs that can disseminate harmful content on social media and even manipulate public opinion. Social bot detection aims to identify bot accounts, which is crucial for maintaining a healthy online ecosystem. However, advances in multimedia technology and smart device prevalence have diversified user-generated social media content, which now encompasses text, images, videos, and more. Effectively leverage multimodal content for robust social bot detection presents a significant research challenge. Furthermore, existing detection methods often overlook the latent social relationships, which we believe can significantly enhance bot detection. To address the issues, in this paper we propose a novel approach for social bot detection that comprehensively leverages users’ multimodal information. Specifically, we first develop an adaptive multimodal fusion mechanism capable of effectively integrating heterogeneous modal information under imbalanced data distributions to obtain more discriminative user representations. Second, we design a latent social relationship mining algorithm that reconstructs more complete social graphs to enhance the objectivity and completeness of user multimodal representations. Finally, on the basis of our proposed multimodal information fusion mechanism and latent social relationship mining algorithm, we design a new social bot detection model. We conduct extensive experiments on the TwiBot-20 dataset, demonstrating superior performance over baseline methods with significant improvements in both detection accuracy and F1-score. Comprehensive ablation studies and dimensionality-reduced visualizations of user representations further validate the critical role of multimodal information and the effectiveness of our proposed model.

  • Robust offloading strategy in VEC with computation uncertainty and imperfect CSI

    by Pengcheng Qian, Liang Wang, Jiamin Fan, Zhenzheng Shi, Mengge Li


  • Semantic Importance-Aware Image Transmission in V2X Networks

    by Anna Cai, Liang Wang, Yaguang Lin, Cong Liu, Pengcheng Qian

    Traditional communication focuses on bit accuracy, while semantic communication improves efficiency by considering the meaning of the data. This article introduces semantic communication to intelligent transportation systems (ITSs), specifically image tasks in vehicle-to-everything (V2X) networks. We propose an adaptive signal-to-noise ratio (SNR) image semantic communication model (ASISC) to address the dynamic nature of V2X environments. An evaluation of the real utility function (ERUF) based on task performance is proposed, which takes into account the semantic transmission rate and the semantic energy consumption. We define the semantic importance scores (SISs) to quantify the complexity of image content. To maximize the ERUF and SIS for image transmission, an optimization problem is formulated to promote high-SIS image transmission in V2X networks. In particular, to address the interpretability challenge of neural networks, we propose an independent univariate approach consisting of a content balance sampling step and an approximate modeling step, to transform the original optimization problem into decoupled power allocation and transmission order subproblems. A ternary search algorithm is used for power allocation, and a distance-based transmission order scheme (DistO) is proposed to give preferential treatment to high-SIS image tasks. Simulation results on the CIFAR10 dataset demonstrate that our method outperforms the benchmark schemes, especially in high-SIS scenarios, and the transmission efficiency is greatly improved. The proposed scheme is easy to implement and exhibits excellent performance in the V2X networks.


  • Robust Information Delivery and Energy Efficiency Maximization in D2D-Based V2X Network

    by Pengcheng Qian, Liang Wang, Zhenzheng Shi, Yaguang Lin, Anna Cai

    Intelligent transportation systems (ITS) are transforming modern mobility, with vehicle-to-everything (V2X) communication emerging as a critical technology for enhancing transportation safety and efficiency. However, the dynamic nature of vehicular networks presents significant challenges, including unreliable channel state information and limited spectrum resources. These limitations can compromise the reliable and low-latency transmission of safety-critical data. To address these challenges, this paper proposes a robust approach for device-to-device (D2D)-based V2X communication networks, focusing on jointly optimizing channel reuse and power allocation to maximize information transmission success rate (SR) and average energy efficiency (AE). A two-step strategy is developed: Firstly, a long-timescale Kuhn-Munkres (LTKM) algorithm is devised to improve channel efficiency through intelligent channel reuse decisions. Secondly, the power allocation problem is modeled as a Markov decision process (MDP) and resolved using a proximal policy optimization (PPO)-based algorithm, enhancing the network’s robustness to time-varying vehicular network scenario. Simulation results demonstrate the effectiveness of our proposed method. Compared to the original Kuhn-Munkres algorithm, the signaling overhead of our approach is reduced by approximately 82%. Furthermore, compared to three benchmark schemes, our approach improves overall performance by approximately 11%, 20%, 33%, and 61%, respectively. Moreover, our approach exhibits more stable performance under different vehicle speeds, which further highlights the robustness of the proposed method.


  • A Topology Switching-Based Load Balancing Routing Scheme With Software-Hardware Separation for Satellite Ground Network

    by Liang Wang, Qiong Bai, Yuan Zhang, Weijia Han, Wenqing Sun

    Low Earth orbit (LEO) satellite network can provide the seamlessly global coverage and relatively low propagation delay, which has been envisioned as a key component of future sixth generation mobile communication. However, the highly dynamic network topology and quite limited communication and computation resources of LEO satellites, alongside with uneven traffic distribution and demand of ground users impose a severe challenge for the routing algorithm design. To tackle this issue, we propose a ground block division and topology switching routing algorithm with software-hardware separation to handle the frequent uploading network topology and routing tables issue and reduce the computational and communication overheads. Furthermore, we estimate the long-term traffic demand in different ground blocks, and calculate the potential traffic demand for each block, and devise the traffic distribution oriented load balancing routing algorithm (TDOLBRA) to avoid bottleneck links and reduce unnecessary overheads. Moreover, considering different Quality of Service (QoS) requirements of ground users’ traffic, we classify the traffic into high priority traffic and regular traffic, and adopt different routing strategies to further promote the network performance and resources utilization. Simulations results on satellite tool kit and optimized network engineering tool demonstrate that the proposed TDOLBRA scheme can have superior performance over three existing methods. To be specific, our proposed TDOLBRA scheme can achieve the best performance with respect to the fairness index, average traffic per link and packet drop rate $P_{\mathrm { loss}}$ . Especially, compared with three benchmark schemes, $P_{\mathrm { loss}}$ under our scheme is reduced by $3.4\times $ , $7.6\times $ , and $13.2\times $ , respectively, when the total traffic is 18000 packets/min.


  • PLP-SSAF: Improved Online Rumor Detection Combining Propagation Link Prediction and Semantic-Structure Adaptive Fusion

    by Chengxin Zhang, Liang Wang, Yaguang Lin, Xiaoming Wang

    The spread of rumors on social networks can diminish public interests, and even pose a threat to social security. The first prerequisite for effectively suppressing rumors in social networks is the precise detection of rumors. However, factors such as hidden social relationships between users, various social contents, and diverse application scenarios in the vast social network bring severe challenges to the timely and accurate detection of rumors. Therefore, for the sake of exploring the influence of the above factors, we propose an adaptive rumor detection model combining propagation link prediction (PLP) and semantic-structure adaptive fusion, PLP and semantic-structure adaptive fusion (PLP-SSAF). First, we investigate in depth the impact of hidden social relationships on the propagation structure of posts. We leverage the hidden social relationships between users, and utilize the graph attention neural network to extract propagation structure features that simultaneously include both current and future propagation structure. Second, in order to overcome regional and cultural differences among users in multidialect environment, we enhance the text in the original dataset and merge it with the source text. We use a language representation model to extract semantic features from the joint text, getting joint semantic features that can more comprehensively represent the text. Finally, we propose a parallel features adaptive fusion mechanism that can dynamically update the weights between propagating structure features and semantic features. This enables the obtained post features to better adapt to rumor detection scenarios in social networks under massive data environments. Extensive experiments on three real-world datasets show that our PLP-SSAF significantly improving rumor detection performance in accuracy and generalization over existing methods, and demonstrates superior rumor detection capabilities at early stages.