TC: Small: Trustworthy Mobile Sensing

Duke University, Durham NC

Investigators

Abstract

Consumer mobile devices such as smartphones have been proposed for use in a wide range of settings, including citizen journalism, citizen science, and domestic eldercare. Re-purposing consumer devices in these domains is appealing because consumer devices offer a low-cost, highly-scalable platform for mobile sensing that can expand coverage and reduce costs. At the same time, these application scenarios raise new data-integrity and privacy challenges. Organizations currently rely on non-technical economic and legal frameworks to establish trust in sensed data: integrity guidelines specify how data can be gathered and handled, and data sources are held accountable through binding legal agreements. Unfortunately, these frameworks are not anonymous and do not scale to the billions of world-wide mobile-phone users. Trustworthy mobile sensing offers a technical way to ensure data integrity without sacrificing anonymity. This project investigates building a platform for trustworthy mobile sensing using Trusted Platform Module (TPM) hardware. Three trusted software subsystems sit on top of TPM functionality: 1) a trusted path for sensor data from sensing hardware to untrusted applications' address spaces, 2) a trusted runtime for transparently monitoring untrusted applications as they execute, and 3) a trusted post-facto analysis engine for comparing untrusted applications' input sensor data to the digital artifacts they generate. Taken together, this combination of hardware and software provides a strong foundation for reasoning about the trustworthiness of anonymous, user-generated content.