Validation and Feasibility of Differentially Private Local Aggregation of Real-Time Data Streams from Resource-Constrained Healthcare IoT Edge Devices

Abstract

Differential privacy (DP) techniques provide important mathematical guarantees of privacy and in particular local DP mechanisms used to protect individual privacy without needing to trust any external entity. However, validation of these techniques is usually carried out using static datasets since IoT devices generating real-time streaming data pose additional difficulties. Hence, current work aims to validate the effectiveness of one such scheme, Privacy-Preserving Endpoint Aggregation (PPEA), on real-time private data obtained from resource-constrained edge devices by measuring utility metrics for the average operation aggregate function. This paper aims to study the feasibility of implementing PPEA for periodic real-time heart rate collection from fitness trackers, which are pervasive IoT devices within the personal healthcare domain capable of recording individual's private data, by considering factors like memory consumption, execution time, and power consumption. We address challenges concerning resource limitations on edge devices regarding lacking out-of-the-box provisions for implementing randomization techniques to achieve DP on streaming data.KeywordsDifferential privacyLocal data aggregationIoT edge devicesFitness trackersReal-time data streamsHeart rate data acquisition