Jinhua Chen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Jinhua Chen has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
- Keping Yu
Hosei University
4 shared publications - Osama Alfarraj
King Saud University
3 shared publications - Mohsen Guizani
Mohamed bin Zayed University of Artificial Intelligence
2 shared publications - Qibin Zhao
School of Automation, Guangdong University of Technology, Guangzhou, China
2 shared publications - Shilong Zhang
Graduate School of Science and Engineering, Hosei University, Tokyo, Japan
2 shared publications - Tong Liu
Graduate School of Science and Engineering, Hosei University, Tokyo, Japan
2 shared publications
Latest Publications
- 29 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 75 (6) , 11607-11620
Joint Resource Allocation and RIS Beamforming for Enhanced Computational Efficiency in Multi-RIS Assisted MEC Networks
Mobile edge computing (MEC) has become a key technique for enhancing wireless devices in terms of computation performance by enabling task offloading to proximate edge servers. However, the performance of MEC networks is often limited by unfavorable wireless propagation conditions, which degrade the reliability of communication links required for efficient task offloading. Reconfigurable intelligent surfaces (RIS) have recently been proposed as a promising solution to mitigate these challenges by intelligently reconfiguring the wireless environment through passive beamforming, where the phase shifts of reflecting elements are adjusted to reshape the propagation channel. Nevertheless, single-RIS deployments generally provide limited signal coverage, which is insufficient to support spatially distributed users. To overcome this limitation, this paper explores a MEC framework supported by multiple RISs, where several surfaces are strategically placed to jointly improve network connectivity and computational efficiency. A joint optimization framework is established to enhance the system-wide computation efficiency by simultaneously optimizing user-RIS association, RIS passive beamforming, and the allocation of communication and computation resources. To tackle this highly non-convex problem, we propose an efficient hierarchical optimization framework, termed hierarchical computational efficiency resource allocation (H-CERA). This framework adopts an outer-loop algorithm based on the Dinkelbach method to transform the original sum-of-ratios objective into a tractable form, and an inner-loop alternating optimization procedure to iteratively update the decoupled variable blocks. Comprehensive simulation experiments confirm that the proposed H-CERA framework is effective, showing substantial improvements in computational efficiency over representative baseline schemes. The results further demonstrate its robustness and scalability under diverse system configurations.
- 29 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 75 (6) , 11607-11620
Joint Resource Allocation and RIS Beamforming for Enhanced Computational Efficiency in Multi-RIS Assisted MEC Networks
Mobile edge computing (MEC) has become a key technique for enhancing wireless devices in terms of computation performance by enabling task offloading to proximate edge servers. However, the performance of MEC networks is often limited by unfavorable wireless propagation conditions, which degrade the reliability of communication links required for efficient task offloading. Reconfigurable intelligent surfaces (RIS) have recently been proposed as a promising solution to mitigate these challenges by intelligently reconfiguring the wireless environment through passive beamforming, where the phase shifts of reflecting elements are adjusted to reshape the propagation channel. Nevertheless, single-RIS deployments generally provide limited signal coverage, which is insufficient to support spatially distributed users. To overcome this limitation, this paper explores a MEC framework supported by multiple RISs, where several surfaces are strategically placed to jointly improve network connectivity and computational efficiency. A joint optimization framework is established to enhance the system-wide computation efficiency by simultaneously optimizing user-RIS association, RIS passive beamforming, and the allocation of communication and computation resources. To tackle this highly non-convex problem, we propose an efficient hierarchical optimization framework, termed hierarchical computational efficiency resource allocation (H-CERA). This framework adopts an outer-loop algorithm based on the Dinkelbach method to transform the original sum-of-ratios objective into a tractable form, and an inner-loop alternating optimization procedure to iteratively update the decoupled variable blocks. Comprehensive simulation experiments confirm that the proposed H-CERA framework is effective, showing substantial improvements in computational efficiency over representative baseline schemes. The results further demonstrate its robustness and scalability under diverse system configurations.
- 20 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (9) , 18129-18142
Intent-Driven DRL-Based Resource Allocation for RIS-Assisted Wireless-Powered IoT Edge Networks
The rapid proliferation of the Industrial Internet of Things (IoT), particularly in smart manufacturing, necessitates intent-driven edge intelligence to provide highly reliable, low-latency computational services for a vast array of connected devices. However, inherent challenges such as the limited computational capacity of IoT devices (IDs) and unpredictable wireless channels hinder the development of such networks. To address these obstacles, this article presents an intent-driven optimization framework that integrates reconfigurable intelligent surfaces (RISs) and wireless power transfer (WPT) into an IoT edge computing architecture. Following the principles of intent-based networking (IBN), our framework targets binary task offloading for energy-constrained devices by translating the specific high-level intent of maximizing the total computational rate into a concrete system-level objective. We formulate this objective as a mixed-integer nonlinear programming (MINLP) problem and propose a deep reinforcement integrated optimization (DRIO) framework to solve it under dynamic network conditions. DRIO effectively decouples the high-dimensional action space, employing a double deep Q-network (DDQN) for coarse RIS phase control, a combination of gradient search and element-wise local search for fine-tuning RIS phases, and a dedicated deep reinforcement learning (DRL) component for managing binary offloading decisions. Simulation results confirm that DRIO surpasses traditional and standard reinforcement learning methods by successfully fulfilling the intent of maximizing computational rate, showing adaptability to dynamic environments and energy harvesting fluctuations. This study enhances IBN for the industrial IoT through AI-powered resource allocation, thereby boosting scalability and efficiency in next-generation communications.
- 20 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (9) , 18129-18142
Intent-Driven DRL-Based Resource Allocation for RIS-Assisted Wireless-Powered IoT Edge Networks
The rapid proliferation of the Industrial Internet of Things (IoT), particularly in smart manufacturing, necessitates intent-driven edge intelligence to provide highly reliable, low-latency computational services for a vast array of connected devices. However, inherent challenges such as the limited computational capacity of IoT devices (IDs) and unpredictable wireless channels hinder the development of such networks. To address these obstacles, this article presents an intent-driven optimization framework that integrates reconfigurable intelligent surfaces (RISs) and wireless power transfer (WPT) into an IoT edge computing architecture. Following the principles of intent-based networking (IBN), our framework targets binary task offloading for energy-constrained devices by translating the specific high-level intent of maximizing the total computational rate into a concrete system-level objective. We formulate this objective as a mixed-integer nonlinear programming (MINLP) problem and propose a deep reinforcement integrated optimization (DRIO) framework to solve it under dynamic network conditions. DRIO effectively decouples the high-dimensional action space, employing a double deep Q-network (DDQN) for coarse RIS phase control, a combination of gradient search and element-wise local search for fine-tuning RIS phases, and a dedicated deep reinforcement learning (DRL) component for managing binary offloading decisions. Simulation results confirm that DRIO surpasses traditional and standard reinforcement learning methods by successfully fulfilling the intent of maximizing computational rate, showing adaptability to dynamic environments and energy harvesting fluctuations. This study enhances IBN for the industrial IoT through AI-powered resource allocation, thereby boosting scalability and efficiency in next-generation communications.
- 16 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 12 (20) , 41349-41364
MCMFL: Monte-Carlo-Dropout-Based Multimodal Federated Learning for Giant Models in 6G Symbiotic Internet of Things
Giant AI models, typically trained and deployed centrally in the cloud, demand significant computational resources, posing privacy risks for the Internet of Things (IoT), particularly in the era of 6G-driven connectivity. Federated Learning (FL) mitigates this by enabling local training and server-side aggregation, fostering 6G Symbiotic IoT while preserving privacy in 6G networks. However, data heterogeneity (DH) in multimodal settings remains a formidable challenge, degrading model performance. While prior studies attribute DH to uneven data distributions, our empirical analysis reveals that hard samples also drive DH, manifesting across both local and global models. To address this, we propose MCMFL, a multimodal FL framework leveraging Monte Carlo dropout to quantify sample uncertainty and identify hard samples. Exploiting 6G’s excellent capabilities, MCMFL optimizes the local loss function and introduces MC-Agg, a robust aggregation algorithm, enhancing the model’s resilience to hard samples. Extensive experiments show that MCMFL demonstrates superior performance, outperforming baseline aggregation methods by up to 5.38% on CIFAR-100, leading local enhancement baselines by 3.14% on TinyImageNet-200 and achieving the highest score of 4.79 in MTBenchmark for LLMs. By shifting the focus from data distribution to sample-level uncertainty, MCMFL provides a novel framework for deploying large AI models via FL in IoT scenarios, mitigating the critical challenge of data heterogeneity and enhancing model robustness.
- 16 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 12 (20) , 41349-41364
MCMFL: Monte-Carlo-Dropout-Based Multimodal Federated Learning for Giant Models in 6G Symbiotic Internet of Things
Giant AI models, typically trained and deployed centrally in the cloud, demand significant computational resources, posing privacy risks for the Internet of Things (IoT), particularly in the era of 6G-driven connectivity. Federated Learning (FL) mitigates this by enabling local training and server-side aggregation, fostering 6G Symbiotic IoT while preserving privacy in 6G networks. However, data heterogeneity (DH) in multimodal settings remains a formidable challenge, degrading model performance. While prior studies attribute DH to uneven data distributions, our empirical analysis reveals that hard samples also drive DH, manifesting across both local and global models. To address this, we propose MCMFL, a multimodal FL framework leveraging Monte Carlo dropout to quantify sample uncertainty and identify hard samples. Exploiting 6G’s excellent capabilities, MCMFL optimizes the local loss function and introduces MC-Agg, a robust aggregation algorithm, enhancing the model’s resilience to hard samples. Extensive experiments show that MCMFL demonstrates superior performance, outperforming baseline aggregation methods by up to 5.38% on CIFAR-100, leading local enhancement baselines by 3.14% on TinyImageNet-200 and achieving the highest score of 4.79 in MTBenchmark for LLMs. By shifting the focus from data distribution to sample-level uncertainty, MCMFL provides a novel framework for deploying large AI models via FL in IoT scenarios, mitigating the critical challenge of data heterogeneity and enhancing model robustness.
- 8 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Social Systems
- Vol. 12 (6) , 5433-5448
A Robust Aggregation of Federated Large Language Models for Multimodal Knowledge Discovery in Computational Social Systems
Amid a rapidly evolving information era, large-scale multimodal knowledge discovery in computational social systems emerges as a key research domain. Large language models (LLMs) play a crucial role in this field, providing contextual understanding and task adaptability. Yet, centralized training of LLM raises privacy concerns. Federated learning (FL) offers a distributed alternative, but it struggles with data heterogeneity and security issues related to model parameters. To this end, we propose a robust aggregation method that leverages the relative total distance of models to improve global model performance in heterogeneous settings, complemented by Cheon-Kim-Kim-Song (CKKS) encryption to secure parameters against parameter stealing without performance loss. Extensive numeric results show our approach excels in LLM testing, scoring 3.74 on MTBenchmark and 8.17 on Vicuna, outperforming state-of-the-art FL methods against data heterogeneity challenges. It also achieves consistent gains on image datasets such as SVHN, CIFAR10, MNIST, TinyImageNet200, and CIFAR100, TinyImageNet200. In summary, our method offers an effective solution for secure multimodal data analysis in computational social systems.
- 8 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Social Systems
- Vol. 12 (6) , 5433-5448
A Robust Aggregation of Federated Large Language Models for Multimodal Knowledge Discovery in Computational Social Systems
Amid a rapidly evolving information era, large-scale multimodal knowledge discovery in computational social systems emerges as a key research domain. Large language models (LLMs) play a crucial role in this field, providing contextual understanding and task adaptability. Yet, centralized training of LLM raises privacy concerns. Federated learning (FL) offers a distributed alternative, but it struggles with data heterogeneity and security issues related to model parameters. To this end, we propose a robust aggregation method that leverages the relative total distance of models to improve global model performance in heterogeneous settings, complemented by Cheon-Kim-Kim-Song (CKKS) encryption to secure parameters against parameter stealing without performance loss. Extensive numeric results show our approach excels in LLM testing, scoring 3.74 on MTBenchmark and 8.17 on Vicuna, outperforming state-of-the-art FL methods against data heterogeneity challenges. It also achieves consistent gains on image datasets such as SVHN, CIFAR10, MNIST, TinyImageNet200, and CIFAR100, TinyImageNet200. In summary, our method offers an effective solution for secure multimodal data analysis in computational social systems.