Collaborative Research: SaTC: CORE: Small: Privately Collecting and Analyzing V2X Data for Urban Traffic Modeling

合作研究：SaTC：核心：小型：私下收集和分析用于城市交通建模的 V2X 数据

• 批准号: 2302689
• 负责人: Yuan Hong
• 金额: $ 29.99万
• 依托单位: University of Connecticut
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-10-01 至 2024-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2302689&HistoricalAwards=false
• 关键词: Collaborative; Research; SaTC; CORE; Small

When widely deployed, Vehicle-to-Everything (V2X) communications in connected vehicles can result in very large-scale and valuable datasets that can be useful for a wide range of transportation safety, mobility, and other related applications. Mandates are being proposed for all new light vehicles to install V2X devices in the near future for such beneficial data collection. A deployable privacy preserving toolkit is critically needed for privately collecting and analyzing V2X data so that the envisioned applications can be fully functional. This project aims at addressing such privacy concerns in practical V2X data collection and analysis for urban traffic modeling, and thus will facilitate the real-world deployment of connected vehicles and V2X systems/applications. Furthermore, this project integrates research results into the curricula at Illinois Institute of Technology, and University of Washington, and provides opportunities for graduate and undergraduate students, especially under-represented and minority students, to participate in cutting-edge research. It also disseminates state-of-the-art privacy preserving techniques into the intelligent transportation and connected vehicles communities.This project develop a series of novel privacy preserving V2X data collection and analysis techniques with provable privacy guarantees. In the first research thrust, novel V2X data collection schemes will be developed to locally perturb V2X data by each vehicle and they will be aggregated for large-scale urban traffic modeling while satisfying the emerging rigorous notion of local differential privacy (LDP). In the second research thrust, novel cryptographic protocols under the secure multiparty computation (MPC) theory will be designed for the infrastructure and vehicles to securely analyze the V2X data for small-scale urban traffic modeling. Such two categories of privacy preserving techniques are expected to fundamentally advance the literature of LDP and MPC (e.g., designing new randomization mechanisms for LDP). The research team will theoretically prove the privacy guarantees for them, and experimentally evaluate their system performance on emulation platforms, as well as deploy them in real-world connected vehicles testbeds.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

当广泛部署时，联网车辆中的车对万物 (V2X) 通信可以产生非常大规模且有价值的数据集，这些数据集可用于广泛的交通安全、移动性和其他相关应用。正在提议要求所有新型轻型车辆在不久的将来安装 V2X 设备，以收集有益的数据。私下收集和分析 V2X 数据迫切需要一个可部署的隐私保护工具包，以便设想的应用程序能够充分发挥作用。该项目旨在解决城市交通建模的实际 V2X 数据收集和分析中的此类隐私问题，从而促进联网车辆和 V2X 系统/应用程序的实际部署。此外，该项目将研究成果融入伊利诺伊理工学院和华盛顿大学的课程中，为研究生和本科生，特别是弱势群体和少数族裔学生提供参与前沿研究的机会。它还将最先进的隐私保护技术传播到智能交通和互联车辆社区。该项目开发了一系列新颖的隐私保护V2X数据收集和分析技术，并具有可证明的隐私保证。在第一个研究重点中，将开发新颖的 V2X 数据收集方案，以本地干扰每辆车的 V2X 数据，并将它们聚合起来以进行大规模城市交通建模，同时满足新兴的严格的本地差分隐私 (LDP) 概念。在第二个研究重点中，将为基础设施和车辆设计安全多方计算（MPC）理论下的新型加密协议，以安全地分析小规模城市交通建模的V2X数据。这两类隐私保护技术有望从根本上推进 LDP 和 MPC 的文献（例如，为 LDP 设计新的随机化机制）。研究团队将从理论上证明其隐私保证，并在仿真平台上实验评估其系统性能，并将其部署在现实世界的联网车辆测试台中。该奖项反映了 NSF 的法定使命，通过评估认为值得支持利用基金会的智力优势和更广泛的影响审查标准。

项目成果

Task-Agnostic Privacy-Preserving Representation Learning for Federated Learning against Attribute Inference Attacks

用于对抗属性推断攻击的联邦学习的任务无关的隐私保护表示学习

• DOI: 10.1609/aaai.v38i10.28965
• 发表时间: 2024-03
• 期刊: Proceedings of the AAAI Conference on Artificial Intelligence
• 作者: Arroyo Arevalo, Caridad;Noorbakhsh, Sayedeh Leila;Dong, Yun;Hong, Yuan;Wang, Binghui
• 通讯作者: Wang, Binghui

WPES '22: 21st Workshop on Privacy in the Electronic Society

WPES 22：第 21 届电子社会隐私研讨会

• DOI: 10.1145/3548606.3563220
• 发表时间: 2022-11
• 期刊: Proceedings of the 2022 ACM SIGSAC Conference on Computer and Communications Security
• 作者: Hong, Yuan;Wang, Lingyu
• 通讯作者: Wang, Lingyu

Poster: Cryptographic Inferences for Video Deep Neural Networks

海报：视频深度神经网络的密码推理

• DOI: 10.1145/3548606.3563543
• 发表时间: 2022-11
• 期刊: Proceedings of the 2022 ACM SIGSAC Conference on Computer and Communications Security (CCS
• 作者: Liu, Bingyu;Wang, Rujia;Ba, Zhongjie;Zhou, Shanglin;Ding, Caiwen;Hong, Yuan
• 通讯作者: Hong, Yuan

Text-CRS: A Generalized Certified Robustness Framework against Textual Adversarial Attacks

Text-CRS：针对文本对抗性攻击的通用认证稳健性框架

• DOI: 10.1109/sp54263.2024.00053
• 发表时间: 2023-07-31
• 期刊: ArXiv
• 作者: Xinyu Zhang;Hanbin Hong;Yuan Hong;Peng Huang;Binghui Wang;Zhongjie Ba;Kui Ren
• 通讯作者: Kui Ren

OpBoost: a vertical federated tree boosting framework based on order-preserving desensitization

OpBoost：基于保序脱敏的垂直联合树提升框架

• DOI: 10.14778/3565816.3565823
• 发表时间: 2022-10
• 期刊: Proceedings of the VLDB Endowment
• 影响因子: 2.5
• 作者: Li, Xiaochen;Hu, Yuke;Liu, Weiran;Feng, Hanwen;Peng, Li;Hong, Yuan;Ren, Kui;Qin, Zhan
• 通讯作者: Qin, Zhan