| 205 | 0 | 286 |
| 下载次数 | 被引频次 | 阅读次数 |
背景知识攻击通过利用先验知识推断用户行为模式，严重威胁轨迹隐私安全。现有轨迹重构方法存在两大局限：一是传统方法未能有效处理语义位置信息泄露问题，且缺乏对含噪数据下深度学习模型的优化，导致轨迹重构精度与语义提取能力不足；二是现有方法泛化能力弱，难以在异构数据集中实现高效重构，制约隐私保护全面性。为此，本研究提出基于语义信息编码的虚假轨迹重构方法（a deep learning-based semantic encoding method for synthetic trajectory reconstruction,DL-SESTR）。该方法结合双向长短时记忆网络（bidirectional long short-term memory,BiLSTM）与注意力机制，捕捉轨迹时空依赖关系并动态筛选关键点以增强抗噪能力；提出兴趣点语义标注算法（pointof-interest semantic annotation algorithm,PSA），通过高效匹配多源兴趣点（point-of-interest,POI）数据提升标注效率；提出基于哈斯图的层次化语义编码算法（Hasse diagram based semantic information encoding algorithm,HDSE），构建语义敏感度权重模型，区分高优先级语义信息与噪声。实验基于T-Drive和GeoLife数据集，验证了模型在密集/稀疏区域、不同隐私预算及昼夜场景下的性能。结果表明：DL-SESTR在隐私保护与数据效用平衡方面显著优于基线方法，Hausdorff距离降低0.3%，动态时间规整（dynamic time warping,DTW）效率提升1.2倍，轨迹平滑度（root mean square,RMS）提高1.18倍；低隐私预算（ε=0.01）下仍保持95%的Euclidean距离缩减率，展现了鲁棒性与泛化能力。
Abstract：Background knowledge attacks pose serious threats to trajectory privacy by exploiting prior knowledge to infer user behavior patterns. Existing trajectory reconstruction methods, however, suffer from two major limitations.First, they fail to address semantic location information leakage effectively, and their deep learning models lack optimization under noisy conditions, leading to inadequate reconstruction accuracy and weak semantic extraction. Second,their poor generalization ability hinders efficient reconstruction across heterogeneous datasets, thereby limiting the comprehensiveness of privacy protection. To address these limitations, this study proposes a deep learning-based semantic encoding method for synthetic trajectory reconstruction(DL-SESTR), a false trajectory reconstruction method based on semantic information encoding. The method integrates a bidirectional long short-term memory network(BiLSTM) with an attention mechanism to capture spatiotemporal dependencies and dynamically identify key trajectory points, thereby improving noise resistance. It also introduces a point-of-interest semantic annotation algorithm(PSA)that matches multi-source point-of-interest(POI) data efficiently to enhance annotation performance. Furthermore, a hierarchical semantic encoding algorithm based on the Hasse diagram(HDSE) is proposed, constructing a semantic sensitivity weight model to distinguish high-priority semantic information from noise. Experiments on the T-Drive and GeoLife datasets evaluated model performance across dense and sparse regions, varying privacy budgets, and dayand-night scenarios. DL-SESTR consistently outperforms baseline methods in balancing privacy protection and data utility: Hausdorff distance is reduced by 0.3%, dynamic time warping(DTW) efficiency improves by 1.2 times,and root mean square(RMS) improves by 1.18 times. Under a low privacy budget( ε = 0.01), the method still achieves a 95% Euclidean distance reduction rate, demonstrating strong robustness and generalization ability.
[1]HEMKUMAR D, RAVICHANDRA S, SOMAYAJULU D V L N. Impact of prior knowledge on privacy leakage in trajectory data publishing[J]. Engineering Science and Technology, an International Journal , 2020, 23(6):1291-1300.
[2]LIU H L, DI W. Application of differential privacy in location trajectory big data[C]//Proceedings of the 2020 International Conference on Intelligent Transportation, Big Data&Smart City(ICITBS), January 11-12, 2020, Vientiane,Laos. New York:IEEE Xplore, 2020:569-573.
[3]PODOLIAKA O, MUSHKATBLAT V, KAPLAN A. Privacy attacks based on correlation of dataset identifiers:assessing the risk[C]//Proceedings of the 2022 IEEE 12th Annual Computing and Communication Workshop and Conference(CCWC), January 26-29, 2022, Las Vegas, NV,USA. New York:IEEE Xplore, 2022:808-815.
[4]CHEN S, FU A M, SHEN J, et al. RNN-DP:a new differential privacy scheme base on recurrent neural network for dynamic trajectory privacy protection[J]. Journal of Network and Computer Applications, 2020, 168:102736.
[5]XIONG J, ZHU H. Real-time trajectory privacy protection based on improved differential privacy method and deep learning model[J]. Journal of Cloud Computing, 2022, 11(1):57.
[6]RAO J M, GAO S, KANG Y H, et al. LSTM-TrajGAN:a deep learning approach to trajectory privacy protection[EB/OL].(2020-06-14)[2025-03-20]. https://arxiv.org/abs/2006.10521.
[7]YU B, YIN H T, ZHU Z X. Spatio-temporal graph convolutional networks:a deep learning framework for traffic forecasting[C]//Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence, July 13-19, 2018, Stockholm, Sweden. Stockholm:International Joint Conferences on Artificial Intelligence Organization,2018:3634-3640.
[8]LIAN J, REN W W, LI L H, et al. PTP-STGCN:pedestrian trajectory prediction based on a spatio-temporal graph convolutional neural network[J]. Applied Intelligence, 2023,53(3):2862-2878.
[9]MALKOV Y A, YASHUNIN D A. Efficient and robust approximate nearest neighbor search using hierarchical navigable small world graphs[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence , 2020, 42(4):824-836.
[10]GUO C, KARRER B, CHAUDHURI K, et al. Bounding training data reconstruction in private(deep)learning[EB/OL].(2022-01-28)[2025-03-20]. https://arxiv.org/abs/2201.12383.
[11]YANG X Y , QIAO D F. Attention-based bidirectional LSTM network for target tracking[C]//Proceedings of the2021 IEEE International Conference on Electronic Technology, Communication and Information(ICETCI), August 27-29, 2021. Changchun, China. New York:IEEE Xplore, 2021:151-156.
[12]DAI L H, LIU J G, JU Z J. Attention mechanism and bidirectional long short-term memory-based real-time gaze tracking[J]. Electronics, 2024, 13(23):4599.
[13]XU Y, PAN Q, WANG Z F, et al. A novel trajectory prediction method based on CNN, BiLSTM, and multihead attention mechanism[J]. Aerospace, 2024, 11(10):822.
[14]AHMADI A M S, SEMNANI S H. Human trajectory prediction using LSTM with Attention mechanism[EB/OL].(2023-09-01)[2025-03-20]. https://arxiv.org/abs/2309.00331.
[15]CHEN J, LI X C, XIAO Y, et al. FRA-LSTM:a vessel trajectory prediction method based on fusion of the forward and reverse sub-network[EB/OL].(2022-01-19)[2025-03-20]. https://arxiv.org/abs/2201.07606.
[16]ZHANG S, WANG L, ZHU M D, et al. A bi-directional LSTM ship trajectory prediction method based on attention mechanism[C]//Proceedings of the 2021 IEEE 5th Advanced Information Technology, Electronic and Automation Control Conference(IAEAC), March 12-14, 2021,Chongqing, China. New York:IEEE Xplore, 2021:1987-1993.
[17]HOCHREITER S , SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8):1735-1780.
[18]CARLINI N, TRAMER F, WALLACE E, et al. Extracting training data from large language models[EB/OL].(2020-12-14)[2025-03-20]. https://arxiv.org/abs/2012.07805.
[19]YUAN J, ZHENG Y, ZHANG C Y, et al. T-drive:driving directions based on taxi trajectories[C]//Proceedings of the18th SIGSPATIAL International Conference on Advances in Geographic Information Systems, November 2-5, 2010,San Jose, California, USA. New York:Association for Computing Machinery, 2010:99-108.
[20]ZHENG Y, XIE X, MA W Y. GeoLife:a collaborative social networking service among user, location and trajectory[J]. IEEE Data Eng Bull, 2010, 33:32-39.
[21]BUCHHOLZ E, ABUADBBA A, WANG S, et al. Reconstruction attack on differential private trajectory protection mechanisms[C]//Proceedings of the 38th Annual Computer Security Applications Conference, December 5-9, 2022,Austin, Texas, USA. New York:Association for Computing Machinery, 2022:279-292.
[22]XU M K, YU L T, SONG Y, et al. GeoDiff:a geometric diffusion model for molecular conformation generation[EB/OL].(2022-03-06)[2025-03-20]. https://arxiv.org/abs/2203.02923.
[23]JIANG K F, SHAO D X, BRESSAN S, et al. Publishing trajectories with differential privacy guarantees[C]//Proceedings of the 25th International Conference on Scientific and Statistical Database Management, July 29-31, 2013,Baltimore, Maryland, USA. New York:Association for Computing Machinery, 2013:1-12.
[24]YUAN J, ZHENG Y, XIE X, et al. T-drive:enhancing driving directions with taxi drivers′intelligence[J]. IEEE Transactions on Knowledge and Data Engineering, 2013,25(1):220-232.
[25]DING Q, SUN B, ZHANG X M. A traffic-light-aware routing protocol based on street connectivity for urban vehicular ad hoc networks[J]. IEEE Communications Letters,2016, 20(8):1635-1638.
[26]MÜLLER M. Dynamic time warping[M]//Information retrieval for music and motion. Berlin, Heidelberg:Springer,2007:69-84.
[27]ZHENG Y. Trajectory data mining:an overview[J]. ACM Transactions on Intelligent Systems and Technology,2015, 6(3):1-41.
基本信息:
DOI：10.12194/j.ntu.20241126001
中图分类号:TP309;TP18
引用信息:
[1]李同鑫,章静,胡昊泽,等.基于深度学习的语义编码虚假轨迹重构方法[J].南通大学学报(自然科学版),2026,25(01):1-13.DOI:10.12194/j.ntu.20241126001.
基金信息:
国家自然科学基金面上项目(62471139); 福建省卫生健康重大科研项目(2021ZD01001); 福建省医疗卫生中青年骨干人才科研培养项目(GY-H-24179); 福建省教育科研单位专项经费项目(2022639); 福建理工大学科研启动基金项目(GY-S24002)
2024-11-26
2024
2025-07-13
2025-07-16
2025
1
2025-07-28
2025-07-28
2025-07-28