Towards utility-aware privacy-preserving sensor data anonymization in distributed IoT

Abstract

With the increased deployment of Internet of Things (IoT) systems, there come multiple privacy concerns regarding how the sensor data is stored and processed. Our research aims to protect user privacy by developing distributed data anonymization techniques that obscure sensitive information in the raw sensor data before it is used by third-party applications. Since users may have different privacy requirements, we propose to leverage meta-learning to allow our model to adapt to various privacy-utility trade-offs. We will assess the efficacy of the proposed techniques on real-world datasets and design defense mechanisms against various adversaries.