Communication-Efficient Differentially Private Parameter Estimation for Distributed Networks and Systems
DOI:
https://doi.org/10.4108/eetsis.13396Keywords:
data privacy, differential privacy, distributed networks and systems, communication-efficient inference, parameter estimation, federated statistical analysis, Gaussian noise mechanismAbstract
In emerging distributed networks and systems, large-scale data are frequently stored across cloud, edge, and multi-center nodes, making centralized statistical analysis infeasible and creating privacy leakage risks during the transmission of gradients, parameters, and intermediate statistics. To address these issues, this paper proposes a communication-efficient differentially private parameter estimation framework for distributed statistical inference. Two privacy-preserving schemes are designed by injecting multivariate Gaussian noise into transmitted first-order gradients and into both initial parameters and gradients. The analysis considers (ε, δ)-differential privacy under calibrated noise and composition accounting. Under the stated regularity, initialization, partition-growth and noise-scale conditions, the estimators remain statistically consistent and retain the convergence order of the non-private one-step estimator. Monte Carlo simulations under ε=0.5 and δ=10−4 show that the private estimators converge to the true parameter values as the number of machines or local sample size increases, while the privacy budget controls the expected privacy-accuracy tradeoff. A real-world case study on the U.S. Airline Database further demonstrates that the relative deviations between privacy-preserving estimators and non-private baselines remain around 2%, and the per-node runtimes of the two private schemes are only 0.0118 s and 0.0123 s, compared with 19.61 s for centralized computation on the sampled dataset. These results indicate that the proposed method can provide rigorous privacy protection, high communication efficiency, and practically acceptable estimation accuracy for privacy-sensitive distributed data analytics.
References
[1] Sweeney L. K-anonymity: a model for protecting privacy. Int J Uncertain Fuzziness Knowl Based Syst. 2002;10(5):557-70. doi:10.1142/S0218488502001648.
[2] de Montjoye YA, Radaelli L, Singh VK, Pentland AS. Unique in the shopping mall: on the reidentifiability of credit card metadata. Science. 2015;347(6221):536-9. doi:10.1126/science.1256297.
[3] Xue J, Yan W. Edge computing communication privacy protection method based on federated learning algorithm. EAI Endorsed Trans Scalable Inf Syst. 2026;12(9). doi:10.4108/eetsis.12254.
[4] Guo Z, Sun Y, Zhang X, Wu L. An efficient privacy-preserving secure aggregation scheme for federated learning with input verification and dropout resistance. EAI Endorsed Trans Scalable Inf Syst. 2026;12(8). doi:10.4108/eetsis.11991.
[5] Sun Y, Cao S. Optimization of urban traffic signal control system on Hadoop platform based on privacy computing. EAI Endorsed Trans Scalable Inf Syst [Internet]. 2026 May 11 [cited 2026 Nov 18];12(10). Available from: https://eudl.eu/doi/10.4108/eetsis.11814.
[6] Li H. A cloud environment security access control scheme based on federated learning and fuzzy logic integration. EAI Endorsed Trans Scalable Inf Syst. 2026;12(9). doi:10.4108/eetsis.11731.
[7] Wang J, Su J. A review of object detection techniques in IoT-based intelligent transportation systems. Comput Mater Contin. 2025;84(1):125-52. doi:10.32604/cmc.2025.064309.
[8] Su J, Dong H, Xia D, Liu AX, Wang B. A zero trust method for tag array authentication of UHF RFID sensing. IEEE Trans Netw. 2026;34:427-41. doi:10.1109/TON.2025.3599016.
[9] Jin Z, Zhang X, Su J, Zhang L, Shen J. Subgraph-driven lightweight federated learning for spatiotemporal cellular traffic prediction. IEEE Trans Netw Serv Manag. 2026;23:1435-48. doi:10.1109/TNSM.2025.3645253.
[10] Jin Z, Yang C, Ye Y, Zhang L, Shen J, Su J. Mobility-aware semi-asynchronous federated learning for vehicular networks. IEEE Trans Veh Technol. 2026;75(2):2001-12. doi:10.1109/TVT.2025.3603978.
[11] Dwork C, McSherry F, Nissim K, Smith A. Calibrating noise to sensitivity in private data analysis. In: Halevi S, Rabin T, editors. Theory of cryptography. Proceedings of the Third Theory of Cryptography Conference (TCC 2006); 2006 Mar 4-7; New York, NY, USA. Berlin: Springer; 2006. p. 265-84. (Lecture Notes in Computer Science; vol. 3876). doi:10.1007/11681878_14.
[12] Dwork C, Kenthapadi K, McSherry F, Mironov I, Naor M. Our data, ourselves: privacy via distributed noise generation. In: Vaudenay S, editor. Advances in cryptology - EUROCRYPT 2006. Proceedings of the 25th International Conference on the Theory and Applications of Cryptographic Techniques; 2006 May 28-Jun 1; St. Petersburg, Russia. Berlin: Springer; 2006. p. 486-503. (Lecture Notes in Computer Science; vol. 4004). doi:10.1007/11761679_29.
[13] Dwork C, Roth A. The algorithmic foundations of differential privacy. Found Trends Theor Comput Sci. 2014;9(3-4):211-407. doi:10.1561/0400000042.
[14] Dinur I, Nissim K. Revealing information while preserving privacy. In: Proceedings of the Twenty-Second ACM SIGMOD-SIGACT-SIGART Symposium on Principles of Database Systems (PODS '03); 2003 Jun 9-12; San Diego, CA, USA. New York (NY): Association for Computing Machinery; 2003. p. 202-10. doi:10.1145/773153.773173.
[15] Dwork C, McSherry F, Talwar K. The price of privacy and the limits of LP decoding. In: Proceedings of the 39th Annual ACM Symposium on Theory of Computing (STOC '07); 2007 Jun 11-13; San Diego, CA, USA. New York (NY): Association for Computing Machinery; 2007. p. 85-94. doi:10.1145/1250790.1250804.
[16] Dwork C, Yekhanin S. New efficient attacks on statistical disclosure control mechanisms. In: Wagner D, editor. Advances in cryptology - CRYPTO 2008. Proceedings of the 28th Annual International Cryptology Conference; 2008 Aug 17-21; Santa Barbara, CA, USA. Berlin: Springer; 2008. p. 469-80. (Lecture Notes in Computer Science; vol. 5157). doi:10.1007/978-3-540-85174-5_26.
[17] Chan THH, Shi E, Song D. Optimal lower bound for differentially private multi-party aggregation. In: Epstein L, Ferragina P, editors. Algorithms - ESA 2012. Proceedings of the 20th Annual European Symposium on Algorithms; 2012 Sep 10-12; Ljubljana, Slovenia. Berlin: Springer; 2012. p. 277-88. (Lecture Notes in Computer Science; vol. 7501). doi:10.1007/978-3-642-33090-2_25.
[18] Sabater C, Bellet A, Ramon J. An accurate, scalable and verifiable protocol for federated differentially private averaging. Mach Learn. 2022;111(11):4249-93. doi:10.1007/s10994-022-06267-9.
[19] Chen WN, Kairouz P, Özgür A. Breaking the communication-privacy-accuracy trilemma. IEEE Trans Inf Theory. 2023;69(2):1261-81. doi:10.1109/TIT.2022.3218772.
[20] Jordan MI, Lee JD, Yang Y. Communication-efficient distributed statistical inference. J Am Stat Assoc. 2019;114(526):668-81. doi:10.1080/01621459.2018.1429274.
[21] Fan J, Guo Y, Wang K. Communication-efficient accurate statistical estimation. J Am Stat Assoc. 2023;118(542):1000-10. doi:10.1080/01621459.2021.1969238.
[22] Zhou X, Chang L, Xu P, Lv S. Communication-efficient and Byzantine-robust distributed learning with statistical guarantee. Pattern Recognit. 2023;137:109312. doi:10.1016/j.patcog.2023.109312.
[23] Fu J, Hong Y, Ling X, Wang L, Ran X, Sun Z, et al. Differentially private federated learning: a systematic review. ACM Comput Surv. 2026;58(11):271. doi:10.1145/3801079.
[24] Hasırcıoğlu B, Gündüz D. Communication efficient private federated learning using dithering. In: ICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP); 2024 Apr 14-19; Seoul, Republic of Korea. Piscataway (NJ): IEEE; 2024. p. 7575-9. doi:10.1109/ICASSP48485.2024.10446222.
[25] Zhang M, Xie Z, Yin L. Private and communication-efficient federated learning based on differentially private sketches. arXiv [Preprint]. 2024 Oct 8 [cited 2026 Nov 18]. arXiv:2410.05733. Available from: https://arxiv.org/abs/2410.05733.
[26] Li Y, Du W, Han L, Zhang Z, Liu T. A communication-efficient, privacy-preserving federated learning algorithm based on two-stage gradient pruning and differentiated differential privacy. Sensors. 2023;23(23):9305. doi:10.3390/s23239305.
[27] Li M, Xiao D, Liang J, Huang H. Communication-efficient and Byzantine-robust differentially private federated learning. IEEE Commun Lett. 2022;26(8):1725-9. doi:10.1109/LCOMM.2022.3180113.
[28] Carletti V, Foggia P, Mazzocca C, Parrella G, Vento M. SoK: gradient inversion attacks in federated learning. In: Bauer L, Pellegrino G, editors. Proceedings of the 34th USENIX Security Symposium (USENIX Security 25) [Internet]; 2025 Aug 13-15; Seattle, WA, USA. Berkeley (CA): USENIX Association; 2025 [cited 2026 Nov 18]. p. 6439-59. Available from: https://www.usenix.org/conference/usenixsecurity25/presentation/carletti
[29] Wang J, Liu S, Li Y. A review of differential privacy in individual data release. Int J Distrib Sens Netw. 2015;11(10):259682. doi:10.1155/2015/259682.
[30] Zhu X, Li F, Wang H. Least-square approximation for a distributed system. J Comput Graph Stat. 2021;30(4):1004-18. doi:10.1080/10618600.2021.1923517.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Yanling Wang

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
This is an open access article distributed under the terms of the CC BY-NC-SA 4.0, which permits copying, redistributing, remixing, transformation, and building upon the material in any medium so long as the original work is properly cited.