CS262A Final Report: Privacy Preserving DataMuling


引用 0|浏览5
We explore the challenges of implementing a privacy-preserving opportunistic network for Internet of Things (IoT) sensor data collection. Opportunistic networks require no fixed infrastructure and allow edge devices to piggy-back messages through “mule” gateway devices, which can be stationary or mobile. While research interest in such networks has waxed and waned over the years, several large-scale commercial deployments have recently been launched, most notably Amazon’s Sidewalk, fueling renewed interest. As these networks become more prevalent, maintaining the privacy of the individuals who participate in them asmules will become increasingly important. In this project, we demonstrate that current implementations of opportunistic backhaul networks leak access patterns through communication metadata, which network providers can leverage to reconstruct location traces. We further argue that opportunistic network and privacy are notmutually exclusive, and suggest some potential directions to strengthen the networks’ privacy properties. Last, we build and evaluate a system that applies the Express anonymous communication system (USENIX Security ’21) in a data muling context to hide identifying metadata, and evaluate its performance compared to a plaintext baseline backhaul system. Given that opportunistic networks are now seeing large-scale commercial deployments, this project serves as a motivation for designing these systems from the ground up to be privacy preserving.
AI 理解论文
Chat Paper