Deep learning-based privacy-preserving framework for synthetic trajectory generation

Kim, Jong Wook; Jang, Beakcheol

doi:10.1016/j.jnca.2022.103459

Cited by 16 publications

(4 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The development of recurrent neural networks has given rise to GRU (Gate recurrent unit), which can effectively deal with the above problems and capture long range dependencies [8] . In order to improve the accuracy of the prediction mechanism, BiGRU improves the GRU, BiGRU network realizes the full understanding of the historical data through the forward and reverse dimensions, which has a great performance improvement compared to the unidirectional GRU network, and the prediction results are more stable its structure [9] . Using BiGRU network as a prediction framework, its model is:…”

Section: Predictive Mechanismsmentioning

confidence: 99%

Personalized trajectory differential privacy protection mechanism based on spatiotemporal correlation prediction

2024

AJCIS

View full text Add to dashboard Cite

Aiming at the problems of user trajectory privacy budget and personalized demand of existing trajectory differential privacy protection technology, a personalized differential privacy protection mechanism based on sliding window and prediction perturbation is proposed. First, based on the road network topology, the sensitive road sections are classified into levels, and the allocation of personalized privacy budget is achieved by customizing the sensitivity of semantic locations. Then, recurrent neural networks and exponential perturbation methods are utilized to predict the perturbed locations that satisfy the differential privacy and temporal correlation requirements, and service similarity is introduced to detect location availability; If successful, the predicted location is directly used instead of the location of the differential perturbation, which reduces the privacy overhead from successive queries and further improves the utilization of the privacy budget. Finally, a w sliding window based trajectory budget allocation mechanism is designed to dynamically adjust the degree of privacy protection for each position in the trajectory according to the privacy needs of the user of the position. Experiments on real datasets show that the method can better achieve the balance between privacy and utility of trajectory data and improve the usability of published data while strictly protecting privacy.

show abstract

Section: Predictive Mechanismsmentioning

confidence: 99%

Personalized trajectory differential privacy protection mechanism based on spatiotemporal correlation prediction

2024

AJCIS

View full text Add to dashboard Cite

show abstract

“…Shin et al [26], on the other hand, addressed the limitations of the GAN model by introducing a classlevel trajectory data generation model using the auxiliary classifier GAN (AC GAN). Kim et al [27] proposed an adversarial autoencoder (AEE) for applying differential privacy. Similarly, Zhang et al [28] introduced LGAN-DP using a GAN framework with differential privacy.…”

Section: Related Workmentioning

confidence: 99%

Generating Labeled Multiple Attribute Trajectory Data With Selective Partial Anonymization Based on Exceptional Conditional Generative Adversarial Network

Song,

Shin,

Ahn

et al. 2023

IEEE Access

View full text Add to dashboard Cite

Trajectory data generated in location-based service environments contain highly sensitive personal information, making them a prime target for privacy attacks. At the same time, however, valuable statistical information can be obtained from such private data. Optimizing this tradeoff between utility and privacy presents a challenge. This study introduces a novel method for partially anonymizing sensitive areas using a conditional generative adversarial network. The proposed method enables the learning of complex spatial, temporal, and categorical features of the selected sensitive area through the utilization of our condition label structure and loss function. In this study, we evaluate and analyze the contents by considering the spatialtemporal characteristics and dividing them into spatial usability and temporal usability. The experimental results demonstrate that the proposed method outperforms related models that employ generative adversarial networks. We achieved high scores in a majority of spatial evaluation items while also discussing the aspects that obtained relatively low scores.

show abstract

“…Trajectory privacy protection techniques have gained significant attention in research due to their capacity to safeguard the privacy of users’ location and behavioral trajectories, becoming an integral component of location privacy protection strategies. There are three fundamental approaches to achieving trajectory privacy: (1) Fake Trajectory [ 4 , 12 ]. This method involves the addition of fabricated trajectory points [ 12 ] or the exchange of two location pairs that are close in time and space [ 4 ] to obfuscate the authentic trajectory data to preserve privacy.…”

Section: Introductionmentioning

confidence: 99%

“…There are three fundamental approaches to achieving trajectory privacy: (1) Fake Trajectory [ 4 , 12 ]. This method involves the addition of fabricated trajectory points [ 12 ] or the exchange of two location pairs that are close in time and space [ 4 ] to obfuscate the authentic trajectory data to preserve privacy. However, this method is susceptible to exploitation by attackers who can discern real from fake trajectory points by scrutinizing time and speed parameters.…”

Section: Introductionmentioning

confidence: 99%

A Privacy-Preserving Trajectory Publishing Method Based on Multi-Dimensional Sub-Trajectory Similarities

Shen,

Wang,

Zhang

2023

Sensors

View full text Add to dashboard Cite

With the popularity of location services and the widespread use of trajectory data, trajectory privacy protection has become a popular research area. k-anonymity technology is a common method for achieving privacy-preserved trajectory publishing. When constructing virtual trajectories, most existing trajectory k-anonymity methods just consider point similarity, which results in a large dummy trajectory space. Suppose there are n similar point sets, each consisting of m points. The size of the space is then mn. Furthermore, to choose suitable k− 1 dummy trajectories for a given real trajectory, these methods need to evaluate the similarity between each trajectory in the space and the real trajectory, leading to a large performance overhead. To address these challenges, this paper proposes a k-anonymity trajectory privacy protection method based on the similarity of sub-trajectories. This method not only considers the multidimensional similarity of points, but also synthetically considers the area between the historic sub-trajectories and the real sub-trajectories to more fully describe the similarity between sub-trajectories. By quantifying the area enclosed by sub-trajectories, we can more accurately capture the spatial relationship between trajectories. Finally, our approach generates k−1 dummy trajectories that are indistinguishable from real trajectories, effectively achieving k-anonymity for a given trajectory. Furthermore, our proposed method utilizes real historic sub-trajectories to generate dummy trajectories, making them more authentic and providing better privacy protection for real trajectories. In comparison to other frequently employed trajectory privacy protection methods, our method has a better privacy protection effect, higher data quality, and better performance.

show abstract

Deep learning-based privacy-preserving framework for synthetic trajectory generation

Cited by 16 publications

References 36 publications

Personalized trajectory differential privacy protection mechanism based on spatiotemporal correlation prediction

Personalized trajectory differential privacy protection mechanism based on spatiotemporal correlation prediction

Generating Labeled Multiple Attribute Trajectory Data With Selective Partial Anonymization Based on Exceptional Conditional Generative Adversarial Network

A Privacy-Preserving Trajectory Publishing Method Based on Multi-Dimensional Sub-Trajectory Similarities

Contact Info

Product

Resources

About