SCIENCE CHINA Information Sciences, Volume 63 , Issue 12 : 220303(2020) https://doi.org/10.1007/s11432-019-2834-x

Generative adversarial networks enhanced location privacy in 5G networks

More info
  • ReceivedOct 21, 2019
  • AcceptedMar 12, 2020
  • PublishedNov 4, 2020



This work was partly supported by JSPS KAKENHI (Grant No. JP19H04105).


[1] Ahmad I, Kumar T, Liyanage M. Overview of 5G Security Challenges and Solutions. IEEE Comm Stand Mag, 2018, 2: 36-43 CrossRef Google Scholar

[2] Domo. Data never sleeps. 2018. https://www.domo.com/solution/data-never-sleeps-6. Google Scholar

[3] Qu Y, Yu S, Gao L. A Hybrid Privacy Protection Scheme in Cyber-Physical Social Networks. IEEE Trans Comput Soc Syst, 2018, 5: 773-784 CrossRef Google Scholar

[4] Ji X, Huang K, Jin L. Overview of 5G security technology. Sci China Inf Sci, 2018, 61: 081301 CrossRef Google Scholar

[5] Finlayson S G, Bowers J D, Ito J. Adversarial attacks on medical machine learning. Science, 2019, 363: 1287-1289 CrossRef ADS Google Scholar

[6] Dwork C. Differential privacy. In: Proceedings of International Colloquium on Automata, Languages, and Programming, Venice, 2006. Google Scholar

[7] Dwork C, Kenthapadi K, McSherry F, et al. Our data, ourselves: privacy via distributed noise generation. In: Proceedings of the 25th Annual International Conference on the Theory and Applications of Cryptographic Techniques, St. Petersburg, 2006. 486--503. Google Scholar

[8] Gruteser M, Grunwald D. Anonymous usage of location-based services through spatial and temporal cloaking. In: Proceedings of the 1st International Conference on Mobile Systems, Applications, and Services, San Francisco, 2003. Google Scholar

[9] Qu Y, Yu S, Zhou W. Privacy of Things: Emerging Challenges and Opportunities in Wireless Internet of Things. IEEE Wireless Commun, 2018, 25: 91-97 CrossRef Google Scholar

[10] Chaudhary R, Kumar N, Zeadally S. Network Service Chaining in Fog and Cloud Computing for the 5G Environment: Data Management and Security Challenges. IEEE Commun Mag, 2017, 55: 114-122 CrossRef Google Scholar

[11] Zeng D, Gu L, Guo S. Joint Optimization of Task Scheduling and Image Placement in Fog Computing Supported Software-Defined Embedded System. IEEE Trans Comput, 2016, 65: 3702-3712 CrossRef Google Scholar

[12] Hashem Eiza M, Ni Q, Shi Q. Secure and Privacy-Aware Cloud-Assisted Video Reporting Service in 5G-Enabled Vehicular Networks. IEEE Trans Veh Technol, 2016, 65: 7868-7881 CrossRef Google Scholar

[13] Yu S. Big Privacy: Challenges and Opportunities of Privacy Study in the Age of Big Data. IEEE Access, 2016, 4: 2751-2763 CrossRef Google Scholar

[14] Samarati P. Protecting respondents identities in microdata release. IEEE Trans Knowl Data Eng, 2001, 13: 1010-1027 CrossRef Google Scholar

[15] Ashwin Machanavajjhala, Daniel Kifer, Johannes Gehrke, and Muthuramakrishnan Venkitasubramaniam. $L$-diversity: Privacy beyond $k$-anonymity. IEEE Transactions on Knowledge and Data Engineering, 1(1), 2007, doi: 10.1109/ICDE.2006.1. Google Scholar

[16] Ninghui Li , Tiancheng Li , Venkatasubramanian S. Closeness: A New Privacy Measure for Data Publishing. IEEE Trans Knowl Data Eng, 2010, 22: 943-956 CrossRef Google Scholar

[17] Qu Y Y, Yu S, Gao L X, et al. Big data set privacy preserving through sensitive attribute-based grouping. In: Proceedings of IEEE International Conference on Communications (ICC), 2017. Google Scholar

[18] Ma L C, Pei Q Q, Qu Y Y, et al. Decentralized privacy-preserving reputation management for mobile crowdsensing. In: Proceedings of International Conference on Security and Privacy in Communication Systems, 2019. 532--548. Google Scholar

[19] Xu C, Zhu L, Liu Y. DP-LTOD: Differential Privacy Latent Trajectory Community Discovering Services over Location-Based Social Networks. IEEE Trans Serv Comput, 2019, : 1-1 CrossRef Google Scholar

[20] Gu B S, Gao L, Wang X. Privacy on the Edge: Customizable Privacy-Preserving Context Sharing in Hierarchical Edge Computing. IEEE Trans Netw Sci Eng, 2019, : 1-1 CrossRef Google Scholar

[21] Youyang Qu, Mohammad Reza Nosouhi, Lei Cui, and Shui Yu. Privacy preservation in smart cities. In Smart Cities Cybersecurity and Privacy, pages 75--88. Elsevier, 2019. doi: 10.1016/B978-0-12-815032-0.00006-8. Google Scholar

[22] Duan X, Wang X. Authentication handover and privacy protection in 5G hetnets using software-defined networking. IEEE Commun Mag, 2015, 53: 28-35 CrossRef Google Scholar

[23] Zhang A, Lin X. Security-Aware and Privacy-Preserving D2D Communications in 5G. IEEE Network, 2017, 31: 70-77 CrossRef Google Scholar

[24] Ni J, Lin X, Shen X S. Efficient and Secure Service-Oriented Authentication Supporting Network Slicing for 5G-Enabled IoT. IEEE J Sel Areas Commun, 2018, 36: 644-657 CrossRef Google Scholar

[25] Liao D, Li H, Sun G. Location and trajectory privacy preservation in 5G-Enabled vehicle social network services. J Network Comput Appl, 2018, 110: 108-118 CrossRef Google Scholar

[26] Qu Y, Pokhrel S R, Garg S. A Blockchained Federated Learning Framework for Cognitive Computing in Industry 4.0 Networks. IEEE Trans Ind Inf, 2020, : 1-1 CrossRef Google Scholar

[27] Qu Y, Gao L, Luan T H. Decentralized Privacy Using Blockchain-Enabled Federated Learning in Fog Computing. IEEE Internet Things J, 2020, 7: 5171-5183 CrossRef Google Scholar

[28] Yu S, Zhou W, Jia W. Discriminating DDoS Attacks from Flash Crowds Using Flow Correlation Coefficient. IEEE Trans Parallel Distrib Syst, 2012, 23: 1073-1080 CrossRef Google Scholar

[29] Yu S, Zhou W, Doss R. Traceback of DDoS Attacks Using Entropy Variations. IEEE Trans Parallel Distrib Syst, 2011, 22: 412-425 CrossRef Google Scholar

[30] Goodfellow I J, Pouget-Abadie J, Mirza M, et al. Generative adversarial nets. In: Proceedings of Annual Conference on Neural Information Processing Systems, Montreal, 2014. 2672--2680. Google Scholar

[31] Arjovsky M, Chintala S, Bottou L. Wasserstein generative adversarial networks. In: Proceedings of the 34th International Conference on Machine Learning, Sydney, 2017. 214--223. Google Scholar

[32] Li C X, Xu T, Zhu J, et al. Triple generative adversarial nets. In: Proceedings of Annual Conference on Neural Information Processing Systems, Long Beach, 2017. 4091--4101. Google Scholar

[33] Song S, Chaudhuri K, Sarwate A D. Stochastic gradient descent with differentially private updates. In: Proceedigns of IEEE Global Conference on Signal and Information Processing, Austin, 2013. 245--248. Google Scholar

[34] Lee J, Kifer D. Concentrated differentially private gradient descent with adaptive per-iteration privacy budget. In: Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, London, 2018. 1656--1665. Google Scholar

[35] Qu Y Y, Yu S, Zhang J W, et al. GAN-DP: generative adversarial net driven differentially privacy-preserving big data publishing. In: Proceedings of IEEE International Conference on Communications (ICC), 2019. Google Scholar

[36] Youyang Qu, Shui Yu, Wanlei Zhou, and Yonghong Tian. Gan-driven personalized spatial-temporal private data sharing in cyber-physical social systems. IEEE Transactions on Network Science and Engineering, 2020. Google Scholar

[37] Dwork C, Feldman V, Hardt M. The reusable holdout: Preserving validity in adaptive data analysis. Science, 2015, 349: 636-638 CrossRef ADS Google Scholar

[38] Shokri R, Shmatikov V. Privacy-preserving deep learning. In: Proceedings of the 22nd ACM SIGSAC Conference on Computer and Communications Security, Denver, 2015. 1310--1321. Google Scholar

[39] Abadi M, Chu A, Goodfellow I J, et al. Deep learning with differential privacy. In: Proceedings of ACM SIGSAC Conference on Computer and Communications Security, Vienna, 2016. 308--318. Google Scholar

[40] Ács G, Melis L, Castelluccia C, et al. Differentially private mixture of generative neural networks. In: Proceedings of IEEE International Conference on Data Mining, New Orleans, 2017. 715--720. Google Scholar

[41] Elshaer H, Boccardi F, Dohler M, et al. Downlink and uplink decoupling: a disruptive architectural design for 5G networks. In: Proceedings of IEEE Global Communications Conference, Austin, 2014. 1798--1803. Google Scholar

[42] Wang Y X, Fienberg S E, Smola A J. Privacy for free: posterior sampling and stochastic gradient monte carlo. In: Proceedings of the 32nd International Conference on Machine Learning, Lille, 2015. 2493--2502. Google Scholar

[43] Huang S W, Lin C T, Chen S P, et al. Auggan: cross domain adaptation with gan-based data augmentation. In: Proceedings of the 15th European Conference, Munich, 2018. 731--744. Google Scholar

[44] Dheeru D, Taniskidou E K. UCI machine learning repository. 2017. Google Scholar

[45] Dwork C. Differential privacy. In: Encyclopedia of Cryptography and Security. 2nd ed. 2011. 338--340. Google Scholar

  • Figure 1

    (Color online) Comparison between raw dataset (a) and generated synthetic dataset (b). The two figures show the correlations and distributions of each attribute before and after processing. The overall statistic features are quite similar, which guarantees high-level data utility.

  • Figure 2

    (Color online) Data utility evaluation. In the two figures, we compare the data utility in two cases where the iteration times are $1000$ and $10000$, respectively. The improvement of accuracy is significant with the increase of the iteration times. (a) $\epsilon~=~3$, it = 1000; (b) $\epsilon~=~1$, it = 10000.

  • Figure 3

    (Color online) Privacy protection evaluation. This figure reveals the privacy protection level upgrades with the decrease of $\epsilon$, which is the index of privacy protection level under differential privacy framework. (a) $\epsilon~=~3$, it = 1000;protect łinebreak (b) $\epsilon~=~1$, it = 10000.

  • Figure 4

    (Color online) Convergence v.s. learning rate. The learning rates are 0.2, 0.3, and 0.6, respectively. In all cases, the proposed model can achieve an convergence while greater learning rates accelerates the convergence.

  • Table 1  

    Table 1Preliminary of the real-wolrd data set

    Property $s$-longitude1 $s$-latitude1 $s$-longitude2 $s$-latitude2
    Count 150.00 150.00 150.00 150.00
    Mean 5.84 3.05 3.76 1.20
    Std 0.83 0.43 1.76 0.76
    Min 4.30 2.00 1.00 0.10
    25% 5.10 2.80 1.60 0.30
    50% 5.80 3.00 4.35 1.30
    75% 6.40 3.30 5.10 1.80
    Max 7.90 4.40 6.90 2.50

    Algorithm 1 Differentially private posterior sample estimator

    Require:Dataset $D$, privacy budget $\epsilon$, log-likelihood function $f(\cdot|\cdot)$ satisfying ${\rm~sup}_{x,\theta}~\|f(x|\theta\|~\leq~B)$, a prior $\pi(\cdot)$.


    Set $\rho~=~{\rm~min}{1,~\frac{\epsilon}{4B}}$;

    Formulate log-likelihood function and the prior $f'(\cdot|\cdot)~=~\rho~f(\cdot|\cdot)$ and $\pi'~=~(\pi(\cdot))^{\rho}$;

    Return $\theta'$.


    Algorithm 2 Differentially private data augmentation

    Require:Sampled dataset $D_r$, size of the original dataset.

    Output:Full-sized differentially private synthetic dataset.

    Establish encoder EN and segmentation loss $\mathfrak{L}_{\rm~seg}$;

    Establish hybrid weight sharing loss $\mathfrak{L}_{w_{D_r}}$ and $\mathfrak{L}_{w_{D_n}}$;

    Establish circle consistence loss $\mathfrak{L}_{\rm~cc}$;

    Establish adversarial learning loss $\mathfrak{L}_{{\rm~GAN}_1}$ and $\mathfrak{L}_{{\rm~GAN}_2}$;

    Integrate all loss to the final objective function $\mathfrak{L}$;

    Return the full-sized differentially private synthetic dataset.