Article
Computer Science, Information Systems
Zhenguo Shi, Qingqing Cheng, J. Andrew Zhang, Richard Yi Da Xu
Summary: This article proposes an innovative scheme, called AFEE-MatNet, for channel state information (CSI)-based human activity recognition (HAR) using deep learning. AFEE-MatNet combines an activity-related feature extraction and enhancement method with a matching network to achieve transferable features and improve recognition performance. The scheme can be directly applied in new/unseen environments without retraining and outperforms existing state-of-the-art HAR methods in terms of accuracy and training time.
IEEE INTERNET OF THINGS JOURNAL
(2022)
Article
Computer Science, Information Systems
Yuanrun Fang, Fu Xiao, Biyun Sheng, Letian Sha, Lijuan Sun
Summary: This paper introduces a cross-scene activity recognition system based on commercial WiFi. By collecting channel state information data and training with bi-directional long short-term memory network, the system achieves accurate recognition and robustness through transfer learning.
FRONTIERS OF COMPUTER SCIENCE
(2022)
Article
Computer Science, Information Systems
Chunjing Xiao, Yue Lei, Chun Liu, Jie Wu
Summary: WiFi channel state information (CSI)-based activity recognition has sparked a lot of research interest due to its wide availability and privacy protection. Existing recognition approaches struggle to generalize beyond the source domain, and few-shot learning-based and data augmentation-based solutions have limitations. To overcome these limitations, we propose a Mean Teacher-based cross-domain human activity recognition framework called WiTeacher. WiTeacher achieves significant gains without requiring any annotation data from the target domain by using label smoothing-based classification loss and sample relation-based consistency regularization.
IEEE INTERNET OF THINGS JOURNAL
(2023)
Article
Engineering, Electrical & Electronic
Changsheng Zhang, Wanguo Jiao
Summary: In this study, a gesture recognition system called ImgFi is proposed, which is based on WiFi channel state information (CSI). By converting CSI to images and utilizing a convolutional neural network (CNN) for recognition, ImgFi achieves high accuracy and low complexity. The tests in different contexts demonstrate that ImgFi achieves a recognition accuracy of 99.5%.
IEEE SENSORS JOURNAL
(2023)
Article
Engineering, Electrical & Electronic
Jieming Yang, Yanming Liu, Zhiying Liu, Yun Wu, Tianyang Li, Yuehua Yang
Summary: Optimizing human activity recognition through WiFi signal analysis can effectively reduce redundancy and improve accuracy. The research results can significantly enhance recognition accuracy and reduce the cost of recognition time.
INTERNATIONAL JOURNAL OF ANTENNAS AND PROPAGATION
(2021)
Article
Engineering, Electrical & Electronic
BeiMing Yan, Wei Cheng, Yong Li, Xiang Gao, HuiMin Liu
Summary: The study introduces a new scheme for joint activity recognition and indoor localization, which utilizes a multi-view fusion strategy and fully leverages the amplitude and phase information of multiple antennas, resulting in significantly improved recognition accuracy for joint tasks.
DIGITAL SIGNAL PROCESSING
(2022)
Article
Computer Science, Information Systems
Chi Lin, Pengfei Wang, Chuanying Ji, Mohammad S. Obaidat, Lei Wang, Guowei Wu, Qiang Zhang
Summary: The ubiquitous and fine-grained features of WiFi signals make it promising for contactless authentication. Existing methods are sensitive to environmental dynamics and over-depend on certain activities. This article presents WiTL, a transfer learning-based contactless authentication system that detects unique human features and removes environment dynamics. The system includes a Height Estimation algorithm based on Angle of Arrival and a transfer learning technology combined with Residual Network and adversarial network. Experiments demonstrate WiTL's high accuracy in multi-scenes and multi-activities identity recognition, surpassing state-of-the-art systems.
ACM TRANSACTIONS ON SENSOR NETWORKS
(2023)
Article
Computer Science, Information Systems
Jin Zhang, Fuxiang Wu, Bo Wei, Qieshi Zhang, Hui Huang, Syed W. Shah, Jun Cheng
Summary: Recent research focuses on utilizing WiFi signals for human activity recognition, facing challenges such as activity inconsistency and subject-specific issues. To address these challenges, a system is proposed that synthesizes activity data and utilizes a novel deep-learning model for small-size WiFi data. Extensive experiments show significant improvement in performance and robustness.
IEEE INTERNET OF THINGS JOURNAL
(2021)
Article
Computer Science, Information Systems
Jin Zhang, Bo Wei, Fuxiang Wu, Limeng Dong, Wen Hu, Salil S. Kanhere, Chengwen Luo, Shui Yu, Jun Cheng
Summary: Research indicates the potential of device-free WiFi sensing for human identification, but inconsistent mirrored patterns in gait monitoring can affect accuracy. The Gate-ID system is proposed as a solution, utilizing theoretical and deep learning models to achieve accurate identification regardless of different walking directions.
IEEE INTERNET OF THINGS JOURNAL
(2021)
Article
Computer Science, Artificial Intelligence
Guiping Lin, Weiwei Jiang, Sicong Xu, Xiaobo Zhou, Xing Guo, Yujun Zhu, Xin He
Summary: In this article, a work on human activity recognition using a smartphone and off-the-shelf WiFi router is presented. The system captures WiFi channel state information (CSI) data and utilizes machine learning models to classify activities. The system achieves a high overall accuracy of 97.25% and shows promising results in recognizing small-scale motions.
IEEE TRANSACTIONS ON HUMAN-MACHINE SYSTEMS
(2023)
Article
Computer Science, Information Systems
Sheng Tan, Yili Ren, Jie Yang, Yingying Chen
Summary: This article surveys the evolution of WiFi sensing systems utilizing commodity devices over the past decade, categorizing them into activity recognition, object sensing, and localization, and highlighting milestone work and techniques adopted in each category. It also presents the challenges faced by existing WiFi sensing systems and comprehensively discusses the future trending of commodity WiFi sensing.
IEEE INTERNET OF THINGS JOURNAL
(2022)
Article
Computer Science, Artificial Intelligence
Santosh Kumar Yadav, Siva Sai, Akshay Gundewar, Heena Rathore, Kamlesh Tiwari, Hari Mohan Pandey, Mohit Mathur
Summary: Human activity recognition is important in various applications, and traditional modalities face limitations such as different illumination and clutter. WiFi CSI-based activity recognition is gaining momentum due to its advantages, and the proposed CSITime model achieves high accuracy on benchmark datasets, surpassing state-of-the-art methods.
Article
Environmental Sciences
Hongqing Liu, Heng Zhang, Jinmei Shi, Xiang Lan, Wenshuai Wang, Xianpeng Wang
Summary: This article proposes an indoor positioning algorithm based on HOSVD, which accurately estimates angle of arrival (AoA) in complex indoor environments with low computational complexity and low localization error rates.
Article
Chemistry, Analytical
Gunsik Lim, Beomseok Oh, Donghyun Kim, Kar-Ann Toh
Summary: In this study, a score-level fusion structure using Wi-Fi channel state information (CSI) signals for human activity recognition is investigated. The fusion provides a convenient, covert, and non-invasive means of recognizing human activity, particularly useful for healthcare monitoring. Experimental results show that the fusion provides good generalization and a shorter learning processing time compared with state-of-the-art networks.
Article
Computer Science, Information Systems
Takashi Nakamura, Mondher Bouazizi, Kohei Yamamoto, Tomoaki Ohtsuki
Summary: The Wi-Fi CSI-based fall detection system has great potential and is not affected by SNR, with a proposed spectrogram-image-based method outperforming conventional methods in different environmental settings.
IEEE INTERNET OF THINGS JOURNAL
(2022)
Review
Transportation
Weiliang Zeng, Miaosen Wu, Peng Chen, Zhiguang Cao, Shengli Xie
Summary: With the advancement of autonomous vehicle technology, online hailing autonomous taxi system is considered to be one of the most popular public transportation services in the future. Recent studies have focused on demand forecasting, ride matching, path planning, relocation, and pricing strategy for shared online hailing and autonomous taxi services. This study conducted a survey based on 141 representative literatures from 1995 to 2022 to understand the latest developments in the key problems of operating autonomous taxi service. The study also discusses how emerging technologies such as internet of vehicles, big data, cloud and edge computing, and blockchain can enhance the autonomous taxi service, and identifies the current research challenges and public concerns or hurdles in adopting autonomous taxi services.
TRANSPORTMETRICA B-TRANSPORT DYNAMICS
(2023)
Article
Computer Science, Artificial Intelligence
Jianhua Xiao, Jingguo Du, Zhiguang Cao, Xingyi Zhang, Yunyun Niu
Summary: This study investigates the electric vehicle routing problem with time windows and mixed backhauls (EVRPTWMB), and proposes a diversity-enhanced memetic algorithm (DEMA) to effectively solve this problem as well as other related problems.
APPLIED SOFT COMPUTING
(2023)
Article
Automation & Control Systems
Wen Song, Xinyang Chen, Qiqiang Li, Zhiguang Cao
Summary: This article proposes a novel DRL method to learn high-quality PDRs for solving the flexible job-shop scheduling problem. The method combines operation selection and machine assignment as a composite decision and utilizes a heterogeneous graph representation to capture complex relationships.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Engineering, Civil
Rongkai Zhang, Cong Zhang, Zhiguang Cao, Wen Song, Puay Siew Tan, Jie Zhang, Bihan Wen, Justin Dauwels
Summary: We propose a manager-worker framework based on deep reinforcement learning to solve a multiple-vehicle Travelling Salesman Problem (TSP) variant with time window and rejections. Experimental results demonstrate that the proposed framework outperforms baselines in terms of solution quality and computation time, achieving competitive performance for unseen larger instances.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2023)
Article
Engineering, Electrical & Electronic
Keyu Wu, Min Wu, Zhenghua Chen, Ruibing Jin, Wei Cui, Zhiguang Cao, Xiaoli Li
Summary: Domain adaptation transfers knowledge from label-rich source domains to label-scarce target domains for generalized learning in new environments. Partial domain adaptation (PDA) extends this concept by considering scenarios where the target label space is a subset of the source label space. This paper proposes a Reinforced Adaptation Network (RAN) that combines deep reinforcement learning with domain adaptation techniques to address the challenging PDA problem. Experimental results show that RAN significantly outperforms existing state-of-the-art methods on three benchmark datasets.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Engineering, Electrical & Electronic
Mingfeng Fan, Yaoxin Wu, Tianjun Liao, Zhiguang Cao, Hongliang Guo, Guillaume Sartoretti, Guohua Wu
Summary: In this paper, a UAV routing problem in the presence of multiple charging stations is studied, aiming to minimize the total distance traveled by the UAV during traffic monitoring. A deep reinforcement learning based method is presented, which incorporates a multi-head heterogeneous attention mechanism for automatically constructing the route and considering energy consumption. Results show that the method outperforms conventional algorithms in terms of solution quality and runtime, and exhibits strong generalized performance on different problem instances.
IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY
(2023)
Article
Automation & Control Systems
Yuan Jiang, Zhiguang Cao, Jie Zhang
Summary: In this article, a method for solving the 3-D bin packing problem using a DRL agent is proposed. The method leverages a multimodal encoder and action representation learning to improve the ability to handle large-scale instances. Additionally, integrating the DRL agent with constraint programming further enhances the solution quality.
IEEE TRANSACTIONS ON CYBERNETICS
(2023)
Review
Engineering, Industrial
Cong Zhang, Yaoxin Wu, Yining Ma, Wen Song, Zhang Le, Zhiguang Cao, Jie Zhang
Summary: An efficient manufacturing system is crucial for a healthy economy. With the rapid development of science and technology, the modern manufacturing system is becoming increasingly complex, posing new challenges. Machine learning technology, particularly deep (reinforcement) learning, has made remarkable progress and this study reviews how it can address problems in manufacturing systems. The study focuses on scheduling, packing, and routing, examining the state-of-the-art research, trends, and future opportunities and challenges.
IET COLLABORATIVE INTELLIGENT MANUFACTURING
(2023)
Article
Automation & Control Systems
Wen Song, Yi Liu, Zhiguang Cao, Yaoxin Wu, Qiqiang Li
Summary: This paper proposes a novel instance-specific algorithm configuration (AC) method based on end-to-end deep graph clustering for Mixed-Integer Programming (MIP) solvers. By representing MIP instances as bipartite graphs, a random walk algorithm is employed to extract raw features with both numerical and structural information. An auto-encoder is then designed to learn dense instance embeddings unsupervisedly, enabling the clustering of heterogeneous instances into homogeneous clusters for training instance-specific configurations.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Jianan Zhou, Yaoxin Wu, Zhiguang Cao, Wen Song, Jie Zhang, Zhenghua Chen
Summary: Dispatching vehicle fleets to serve flights is a challenging task in airport ground handling due to the growth of flights. This study presents a learning assisted large neighborhood search method to effectively schedule multiple types of operations for a large number of flights. Experimental results show that the proposed method outperforms state-of-the-art methods and demonstrates versatility and scalability.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2023)
Article
Engineering, Electrical & Electronic
Yinqiu Xia, Chengpu Yu, Chaoyang Jiang
Summary: This article studies the problem of distributed localization for a sensor network using noisy distance measurements. A coordinate descent scheme-based method is developed for solving the localization problem, which guarantees convergence and can reach the globally optimal solution. Simulation examples demonstrate the effectiveness of the proposed method in both noise-free and noisy scenarios.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Article
Computer Science, Artificial Intelligence
Hongliang Guo, Qihang Peng, Zhiguang Cao, Yaochu Jin
Summary: This article proposes a unified algorithm named DRL-Searcher for the multirobot efficient search (MuRES) problem in a nonadversarial moving target scenario. DRL-Searcher utilizes distributional reinforcement learning (DRL) to evaluate and improve the search policy's return distribution for both minimizing capture time and maximizing capture probability objectives. The algorithm is further adapted for target search without real-time location information, and a recency reward is introduced for implicit coordination among multiple robots. Comparative simulations and real-world experiments demonstrate the superior performance of DRL-Searcher.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Engineering, Civil
Yaoxin Wu, Jianan Zhou, Yunwen Xia, Xianli Zhang, Zhiguang Cao, Jie Zhang
Summary: This study aims to improve the solution quality and computation efficiency for airport ground handling (AGH). It models AGH as a multiple-fleet vehicle routing problem (VRP) and proposes a neural method to construct routing solutions for sub-problems. Extensive experiments show that this method outperforms classic meta-heuristics and specialized methods for AGH.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2023)
Review
Mathematical & Computational Biology
Kai Cheng, Lixia Li, Yanmin Du, Jiangtao Wang, Zhenghua Chen, Jian Liu, Xiangsheng Zhang, Lin Dong, Yuanyuan Shen, Zhenlin Yang
Summary: Percutaneous puncture is a common medical procedure that can benefit from image guidance and surgical robot-assistance. These technologies have shown potential in improving accuracy and precision of percutaneous procedures, reducing radiation exposure, and lowering complication risks. However, challenges such as integration, perception, and needle insertion deviation need further research to optimize the utilization of these technologies in clinical practice.
MATHEMATICAL BIOSCIENCES AND ENGINEERING
(2023)
Article
Automation & Control Systems
Wen Song, Nan Mi, Qiqiang Li, Jing Zhuang, Zhiguang Cao
Summary: This paper proposes a novel deep reinforcement learning (DRL) method for solving the Stochastic Economic Lot Scheduling Problem (SELSP). The method effectively extracts useful features from raw state information using a self-attention mechanism and is flexible in handling different numbers of products. Experimental results show that the proposed method outperforms recent DRL methods and state-of-the-art hyper-heuristics.
IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING
(2023)