Article
Engineering, Civil
Chuan Xu, Yingyi Ding, Chao Chen, Yong Ding, Wei Zhou, Sheng Wen
Summary: This paper proposes a personalized location privacy protection scheme based on differential privacy to protect the privacy of location-based services in vehicular networks. By establishing a utility model and a sensitivity distance index, differentiated protection for users' different locations is achieved. An optimal solution for false location is obtained through a multi-objective optimization model.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2023)
Article
Computer Science, Information Systems
Hui Wang, Shangqing Cai, Peiqian Liu, Jingyi Zhang, Zihao Shen, Kun Liu
Summary: With the continuous development of intelligent transportation, researchers propose a traffic statistics publication mechanism with differential privacy and a spatial-temporal graph attention network (DPSTGAT) to address the issues of poor prediction accuracy and reduced data utility caused by the lack of attention to spatial temporal correlations and unreasonable privacy budget allocations. The mechanism includes components such as an adjacency matrix based on equivalent distance, a multistep prediction model based on an STGAT, and a combination of pre-allocation and adaptive allocation method for privacy budget allocation. Experimental results show that the proposed scheme outperforms existing methods.
INFORMATION SCIENCES
(2023)
Article
Computer Science, Artificial Intelligence
Chun Liu, Youliang Tian, Jinchuan Tang, Shuping Dang, Gaojie Chen
Summary: This paper proposes a novel LDP-FL framework under multi-privacy regimes, achieving efficient model training in multiple privacy domains. By using maximum likelihood estimation to compute unbiased global gradients and designing two dynamic privacy budget allocation approaches, the efficiency of model training is improved. Additionally, a layered dimension selection strategy is proposed to avoid utility loss caused by perturbing high-dimensional local gradients in traditional methods.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Jiawei Yang, Shuhong Chen, Guojun Wang, Zijia Wang, Zhiyong Jie, Muhammad Arif
Summary: This paper proposes a gradient compression federated learning framework based on adaptive local differential privacy budget allocation (GFL-ALDPA) to reduce the loss of privacy budget and the amount of model noise, and improve model accuracy. By assigning different privacy budgets to different communication rounds during training, it maximizes the limited privacy budget and achieves a better trade-off between privacy preservation, communication efficiency, and model accuracy.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Mathematics
Peng Qin, Lina Wang
Summary: The proposed personalized location privacy protection scheme (PPVC) based on differential privacy can meet users' service demands effectively protect their privacy. By analyzing the utility and privacy impact of recommended routes, integrating users' privacy preferences, assigning appropriate privacy budgets to users, and generating service request locations with the highest utility.
Article
Computer Science, Information Systems
Shahriar Badsha, Xun Yi, Ibrahim Khalil, Dongxi Liu, Surya Nepal, Elisa Bertino, Kwok Yan Lam
Summary: The personalized Web service recommendation based on Quality of Service (QoS) is gaining popularity, with collaborative filtering techniques contributing to high accuracy predictions. User location is also an important factor, and privacy-preserving protocols can ensure secure and practical recommendations without disclosing private information.
IEEE TRANSACTIONS ON SERVICES COMPUTING
(2021)
Article
Computer Science, Information Systems
Jianhao Wei, Yaping Lin, Xin Yao, Jin Zhang, Xinbo Liu
Summary: This article introduces a genetic matching scheme based on differential privacy technology to protect genetic data privacy and achieve effective genetic matching. The scheme constructs noisy published and query sequences using differential privacy algorithms, and calculates the longest common subsequence through a dynamic programming algorithm to achieve matching results.
IEEE TRANSACTIONS ON EMERGING TOPICS IN COMPUTING
(2021)
Article
Computer Science, Information Systems
Xutong Jiang, Yuhu Sun, Bowen Liu, Wanchun Dou
Summary: A combinatorial double auction scheme based on differential privacy is proposed to protect the security and privacy of multi-resource allocation in the trading market.
COMPUTER COMMUNICATIONS
(2022)
Article
Engineering, Manufacturing
Weiwei Chen, Siyang Gao, Wenjie Chen, Jianzhong Du
Summary: This paper focuses on a class of resource allocation problems in service systems, where the service-level objective and constraints are in the form of probabilistic measures. A generalized resource allocation model and an optimal computing budget allocation formulation are proposed to solve the resource allocation problem. The effectiveness of the proposed algorithm for solving the problem via simulation is demonstrated through numerical experiments and a case study.
PRODUCTION AND OPERATIONS MANAGEMENT
(2023)
Article
Engineering, Electrical & Electronic
Ting Bao, Lei Xu, Liehuang Zhu, Lihong Wang, Tielei Li
Summary: The development of vehicle positioning technologies has enabled in-vehicle recommendation systems, with one common form being the successive POI recommendation. This method helps users choose places to visit but raises privacy concerns due to raw check-in data collection. To address this issue, a recommendation framework using local differential privacy and three influence factors has been proposed to provide strong privacy protection for users.
IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY
(2021)
Article
Computer Science, Information Systems
Jiandong Wang, Hao Liu, Xuewen Dong, Yulong Shen, Xinghui Zhu, Bin Wang, Feng Li
Summary: This article proposes a double MCS auction mechanism with a personalized location privacy incentive, which can meet the competition requirement of task requesters and the task preference variance of workers. The mechanism introduces the concept of privacy budget, allowing workers to decide how much location information to disclose to the platform for personalized location privacy protection. In addition, each worker can offer multiple bids for interested tasks and perform a subset of tasks in a bid if wins. Experimental results validate that the mechanism satisfies budget balance, individual rationality, and 2-D-truthfulness.
IEEE INTERNET OF THINGS JOURNAL
(2023)
Article
Computer Science, Information Systems
Ben Niu, Qinghua Li, Hanyi Wang, Guohong Cao, Fenghua Li, Hui Li
Summary: This paper discusses how to utilize multiple single privacy preserving mechanisms (PPMs) for location privacy protection in different scenarios, and proposes a general framework called SmartGuard that dynamically selects the best privacy preservation strategy for a user based on her preferences and the current status of her mobile device. Evaluation results show that the proposed solution outperforms existing PPMs under various scenarios.
IEEE TRANSACTIONS ON MOBILE COMPUTING
(2022)
Article
Engineering, Civil
Chenxi Chen, Xianbiao Hu, Yang Li, Qing Tang
Summary: This paper presents an optimal privacy budget allocation algorithm for publishing privacy-preserving trajectory data. The algorithm utilizes a prefix tree structure to store smart card trajectory data and develops a query probability model to measure the probability of a trajectory location pair being queried. The Lagrangian relaxation method is used to determine the optimal privacy budget values, and the algorithm achieves high data utility and computational efficiency.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2023)
Article
Computer Science, Theory & Methods
Lei Wu, Chengyi Qin, Zihui Xu, Yunguo Guan, Rongxing Lu
Summary: The prevalence of mobile Internet, smart terminal devices, and GPS positioning technology has led to a large amount of trajectory data that location-based applications can use. However, without extra protection, delivering location-based services (LBSs) based on trajectories may expose user's personal information and social ties. Existing works on differential privacy for trajectory correlation focus on single user trajectory correlation and do not consider privacy protection for correlation among multiple users. To address these challenges, we propose a trajectory correlation privacy-preserving mechanism (TCPP) that fulfills differential privacy. Our mechanism filters out correlated trajectories using Euclidean distance, applies Kalman filter for high availability dataset generation, and uses a customized privacy budget allocation strategy for preserving trajectory correlation when publishing trajectories. Rigid security analysis and experimental results on real-world datasets show the effectiveness and advantages of our proposed mechanism in preserving trajectory correlation privacy.
IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY
(2023)
Article
Engineering, Multidisciplinary
Mingyue Zhang, Junlong Zhou, Gongxuan Zhang, Lei Cui, Tian Gao, Shui Yu
Summary: This paper presents a fine-grained personalized differential privacy data publishing scheme (APDP) for social networks. The scheme defines privacy protection levels based on attribute values and maps them to the amount of noise required to add using the TOPSIS method. Additionally, access control is integrated with differential privacy to prevent illegal data downloads. Theoretical analysis and simulations demonstrate that APDP achieves efficient personalized differential privacy data publishing with reasonable data utility.
IEEE TRANSACTIONS ON NETWORK SCIENCE AND ENGINEERING
(2023)
Article
Computer Science, Information Systems
Yucheng Dong, Qin Ran, Xiangrui Chao, Congcong Li, Shui Yu
Summary: In this article, we propose a continual personalized individual semantics learning model to support consensus-reaching in large-scale linguistic group decision making. The model derives personalized numerical scales from linguistic preference data, performs clustering ensemble method for group division and consensus management, and demonstrates its effectiveness through a case study on intelligent route optimization.
ACM TRANSACTIONS ON INTERNET TECHNOLOGY
(2023)
Article
Computer Science, Theory & Methods
Zhiyi Tian, Lei Cui, Jie Liang, Shui Yu
Summary: The prosperity of machine learning has led to an increase in attacks on the training process, with poisoning attacks emerging as a significant threat. Defending against these attacks is challenging, and a systematic review from a unified perspective is lacking. This survey provides a comprehensive overview of poisoning attacks and countermeasures in both centralized and federated learning, categorizing attack methods based on goals and analyzing their differences and connections. Countermeasures in different learning frameworks are presented, along with a discussion of the feasibility of poisoning attacks and potential research directions.
ACM COMPUTING SURVEYS
(2023)
Article
Automation & Control Systems
Laizhong Cui, Jiating Ma, Yipeng Zhou, Shui Yu
Summary: This study proposes a novel differentially private federated learning algorithm with sparse responses (DPFL-SR), which reduces the privacy budget consumption in each global iteration by applying the sparse vector technique. Experimental results demonstrate that DPFL-SR achieves higher model accuracy and privacy protection level in IIoT systems.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Computer Science, Theory & Methods
Zhuangzhuang Zhang, Libing Wu, Chuanguo Ma, Jianxin Li, Jing Wang, Qian Wang, Shui Yu
Summary: Nowadays, edge computing service providers aim to enhance their models by utilizing the computational power and data of edge nodes without data transmission. The existing privacy-preserving federated learning faces challenges due to the complexity of cryptographic algorithms, the lack of Byzantine robustness while maintaining data privacy, and the limited computing power of edge nodes. To address these issues, a lightweight and secure federated learning scheme LSFL is proposed, which combines privacy preservation and Byzantine robustness. The scheme employs a Lightweight Two-Server Secure Aggregation protocol to ensure secure model aggregation and protect data privacy from Byzantine nodes.
IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY
(2023)
Article
Engineering, Electrical & Electronic
Yuepeng Li, Deze Zeng, Lin Gu, Andong Zhu, Quan Chen, Shui Yu
Summary: This article addresses the security issues of offloading tasks to edge servers and proposes a Priority-aware Secure Task Offloading (PASTO) algorithm based on TrustZone. Experimental results show that PASTO effectively reduces the total task completion time compared to other approaches.
IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY
(2023)
Article
Computer Science, Information Systems
Aleteng Tian, Bohao Feng, Huachun Zhou, Yunxue Huang, Keshav Sood, Shui Yu, Hongke Zhang
Summary: In this paper, an efficient cooperative caching (FDDL) framework is proposed to address key issues in mobile edge networks. Machine learning techniques are used to improve content placement and reduce computation complexity and communication costs. The framework includes a cache admission algorithm, a lightweight eviction algorithm for fine-grained replacements, and a federated learning-based parameter sharing mechanism. Experimental results show that the proposed FDDL achieves higher cache hit ratio and traffic offloading rate, and effectively reduces communication costs and training time.
IEEE TRANSACTIONS ON NETWORK AND SERVICE MANAGEMENT
(2023)
Article
Computer Science, Information Systems
Kaiyue Zhang, Zipei Fan, Xuan Song, Shui Yu
Summary: This article proposes a DLGMP algorithm based on deep leakage from gradients and mobility prior knowledge to solve the problem of trajectory data attacks. The algorithm utilizes spatiotemporal structural information as prior mobility knowledge, greatly reducing the difficulty of recovery, and improves the accuracy and reasonableness of trajectory recovery by adding an easily extensible regularization term and an adversarial loss of Wasserstein GAN.
IEEE INTERNET OF THINGS JOURNAL
(2023)
Article
Computer Science, Information Systems
Tian Liu, Jun Xia, Zhiwei Ling, Xin Fu, Shui Yu, Mingsong Chen
Summary: This article presents a distillation-based federated learning (DFL) method that efficiently and accurately handles federated learning for AIoT applications by using knowledge distillation and local model gradients for aggregation and dispatching.
IEEE INTERNET OF THINGS JOURNAL
(2023)
Article
Computer Science, Information Systems
Chenhan Zhang, Shuyu Zhang, Xiexin Zou, Shui Yu, James J. Q. Yu
Summary: This article proposes a network partitioning approach to improve the performance of graph convolutional network-based predictors on large-scale transportation networks. The approach uses both topological features and traffic speed observations for partitioning, and employs a data-parallel training strategy for parallel training. Case studies on real-world datasets show that the proposed approach can improve the accuracy and training efficiency of the predictors.
IEEE INTERNET OF THINGS JOURNAL
(2023)
Article
Computer Science, Artificial Intelligence
Shigen Shen, Xiaoping Wu, Panjun Sun, Haiping Zhou, Zongda Wu, Shui Yu
Summary: This study proposes a signaling game approach for privacy preservation in edge-computing-based IoT networks. It addresses the issue of malicious IoT nodes requesting private data from an IoT cloud storage system across edge nodes. The optimal privacy preservation strategies for edge nodes are derived and a signaling Q-learning algorithm is designed to achieve convergent equilibrium and game parameters. Simulation results show that the proposed algorithm effectively decreases the optimal probability of malicious requests, enhancing privacy preservation in edge-computing-based IoT cloud storage systems.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Guorui Li, Yajun Wu, Cong Wang, Sancheng Peng, Jianwei Niu, Shui Yu
Summary: In this article, a similarity-based relevance vector machine (SRVM) is proposed to address the challenges of insufficient failure data and low confidence in remaining useful lifetime (RUL) prediction results. The relationship among latent variables is learned adaptively through similarity computations to fully utilize the limited degradation data, and the internal variables in SRVM are treated as time-varying variables and re-estimated dynamically to provide reliable confidence for RUL prediction. Experimental results demonstrate that SRVM achieves higher prediction accuracy compared to other baseline methods.
IEEE INTELLIGENT SYSTEMS
(2023)
Article
Computer Science, Hardware & Architecture
Weibei Fan, Fu Xiao, Hui Cai, Xiaobai Chen, Shui Yu
Summary: This paper studies fault-tolerant routings in BCube data center network. A fault-tolerant routing algorithm based on node disjoint multi-paths is proposed, which has stronger fault tolerance. An effective fault-tolerant routing algorithm based on routing capabilities for BCube is investigated, which has higher fault tolerance and success rate. An adaptive path finding algorithm is presented to establish virtual links in BCube, which can shorten the diameter. Extensive simulations show that the proposed routing scheme outperforms existing algorithms, achieving significant improvements in throughput, packet arrival rate, and latency.
IEEE TRANSACTIONS ON COMPUTERS
(2023)
Article
Engineering, Electrical & Electronic
Youxiang Duan, Yuxi Lu, Shigen Shen, Shui Yu, Peiying Zhang, Wei Zhang, Kostromitin Konstantin Igorevich
Summary: This paper proposes an SFC path optimization strategy based on the clustering of network functional layouts. By designing a topological network identification algorithm and a multi-headed attention mechanism, different types of SFC layouts can be optimized in a targeted manner. The experimental results demonstrate the effectiveness of this strategy in optimizing SFC paths and its ability to optimize service function chain paths in future complex network situations.
IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY
(2023)
Article
Telecommunications
Jiale Zhang, Yue Liu, Di Wu, Shuai Lou, Bing Chen, Shui Yu
Summary: This article introduces the importance of federated learning for edge computing and the challenges of privacy protection. To address the shortcomings of traditional privacy-preserving federated learning schemes, a verifiable privacy-preserving federated learning scheme is proposed. It combines the Distributed Selective Stochastic Gradient Descent (DSSGD) method with the Paillier homomorphic cryptosystem to achieve distributed encryption functionality and presents an online/offline signature method for lightweight gradient integrity verification.
DIGITAL COMMUNICATIONS AND NETWORKS
(2023)
Article
Computer Science, Theory & Methods
Shigen Shen, Lanlan Xie, Yanchun Zhang, Guowen Wu, Hong Zhang, Shui Yu
Summary: This paper proposes a two-layer malware spread-patch model based on IIoT and designs a new algorithm suitable for suppressing the spread of malware. The effectiveness of the model and algorithm is verified through in-depth analysis and numerous comparative experiments.
IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY
(2023)