A Noise Optimization Mechanism for Geographical Privacy Protection Based on Adaptive Adjustment of k-Metric Sensitivity
DOI:
https://doi.org/10.54097/fh385s36Keywords:
Differential Privacy, k-Metric Sensitivity, Noise Optimization, Fuzzy Logic Controller, Location-Based Services (LBS)Abstract
Various location-based services are widely used today, so location privacy protection is an important topic. Differential privacy has solid theoretical support. However, current methods using k-metric sensitivity will suddenly change noise levels at critical thresholds. This makes service accuracy unstable and brings bad experience to users. To fix this problem, this paper designs an adaptive noise optimization method for k-metric sensitivity. It uses smooth functions and fuzzy logic controllers instead of the old switch mode. The noise amount changes slowly with sensitivity values. We run experiments on the Geolife dataset. Compared with the classic PTR framework, our method removes unstable jumps of location outputs. For moving users, the NRMSE error drops by around 20%. The QoE score, which measures overall user experience, rises by 12% to 18%. The test results show that this adaptive noise optimization method can balance privacy protection and data quality. It provides steady location services for users. This work is helpful for both theoretical research and practical application of geographic privacy protection.
Downloads
References
[1] Qian, K., & Li, X. (2023). A dual privacy protection scheme for LBS users based on trajectory similarity. Computer Simulation, 40(11), 459–465.
[2] Wang, H. (2024). Research on network security and privacy protection technologies in the context of big data. Communications Power Technology, 41(9), 148–150.
[3] Liu, Y., Chen, H., Liu, Y., Zhao, D., & Li, C. (2022). Privacy attacks and defense techniques on graph data. Chinese Journal of Computers, 45(4), 702–734.
[4] Zhang, X., Yang, H., Li, Z., He, F., Gai, J., & Bao, J. (2021). A differential private location privacy protection method integrating semantic and geographical information. Computer Science, 48(8), 300–308.
[5] Zhou, N., Long, S., & Liu, H. (2023). Graph publication via dK-sequence synthesis satisfying differential privacy. Journal of Computer Applications, 40(5), 1528–1534.
[6] Wu, J., & Li, X. (2024). A piecewise clipping federated learning algorithm based on differential privacy. Journal of Computer Applications, 41(5), 1532–1537.
[7] Wang, C., Wang, L., & Yang, H. (2023). An improved K-means algorithm based on privacy preservation. Ordnance Automation, 42(12), 38–45.
[8] Zhou, Z., & Qian, X. (2022). A differential privacy protection algorithm for deep neural networks. Journal of Electronics & Information Technology, 44(5), 1773–1781.
[9] Zhang, J., Liu, D., Shen, Z., Wang, H., & Liu, P. (2023). The KSDP scheme for trajectory data publication. Journal of Shenzhen University (Science and Technology Edition), 40(2), 236–243.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Frontiers in Computing and Intelligent Systems
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
