IMR: MM-1B: Longitudinal End-device based Performance Measurement of Cellular Networks with Provable Privacy

蜂窝网络提供了随时随地访问互联网的便利。测量和改进蜂窝网络的性能对于网络提供商、最终用户、内容提供商和监管机构非常重要。虽然蜂窝网络提供商可以直接测量他们的网络，但他们越来越多地将测量外包给第三方移动分析公司，这些公司直接从最终用户移动设备收集测量结果，以进行可扩展的、低成本的长期和广域测量。然而，现有的基于移动设备的测量平台有两个主要限制。首先，它们不向最终用户提供可证明的隐私保障。其次，它们没有根据设备的位置协调设备之间的测量，这可能会导致测量偏差或浪费资源。该项目的创新之处在于为蜂窝网络的纵向协调测量设计了创新的架构和技术，同时为最终用户提供了可证明的隐私。可证明的隐私基于新兴的本地差异隐私(LDP)模型，在该模型下，终端设备在位置信息离开设备之前扰乱位置信息，因此实际位置永远不会超出终端设备的范围。基于扰动的位置数据，终端设备处的测量被调度和协调以实现高效的资源使用。该项目更广泛的意义和重要性在于提高基于移动设备的数据收集中的隐私意识，在研究中招募未被充分代表的学生，以及与行业合作。该项目有三个主要贡献。首先，它提出了一种基于LDP的放大技术，用于从终端设备收集高精度的蜂窝网络测量，同时为个人用户提供可证明的隐私。其次，提出了一种基于优化的测量调度框架来协调移动设备上的测量，以节省资源使用，同时激励测量。第三，它开发了一个数据驱动的模拟工具包，帮助实践者采用测量框架。研究团队进一步开发了一个原型系统，并将其用于进行用户研究，以进一步验证和改进该系统。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

Cellular networks provide convenient access to the Internet anytime and anywhere. Measuring and improving the performance of cellular networks is important to network providers, end users, content providers, and regulators. While cellular network providers can directly measure their networks, they increasingly outsource the measurements to third-party mobile analytics companies, which collect measurements directly from end-user mobile devices for scalable, low-cost, long-term, and wide-area measurements. Existing mobile-device based measurement platforms, however, have two major limitations. First, they do not provide provable privacy guarantees to end users. Second, they do not coordinate measurements across the devices based on their locations, which can lead to biased measurements or wasted resources. This project’s novelties are in designing innovative architecture and techniques for longitudinal coordinated measurements of cellular networks, while providing provable privacy to end users. The provable privacy is based on the emerging local differential privacy (LDP) model, under which end devices perturb the location information before it leaves the devices, and hence the actual locations are never known beyond the end devices. Based on perturbed location data, the measurements at end devices are scheduled and coordinated to achieve efficient resource usage. The project's broader significance and importance are in raising awareness in privacy in mobile-device based data collection, recruiting underrepresented students in research, and collaborating with industry.This project makes three main contributions. First, it proposes an amplified LDP based technique for collecting cellular network measurements from end devices with high accuracy, while providing provable privacy to individual users. Second, it proposes an optimization-based measurement scheduling framework to coordinate the measurements at the mobile devices to conserve resource usage, while incentivizing measurements. Third, it develops a data-driven simulation toolkit that assists practitioners to adopt the measurement framework. The research team further develops a prototype system and uses it to conduct a user study to further validate and improve the system.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.