Non-Invasive User Tracking via Passive Sensing

Abstract

A large-scale sensing infrastructure can collect ample data to benefit many real-world applications. One promising application scenario is building management. However, exposure of the sensor data potentially reveals private details about building users. In this paper, we investigate indoor location privacy as a motivating example to manifest potential privacy risks in smart buildings. We apply inference techniques to reconstruct users' location traces from room-level occupancy data. Unlike other types of surveillance that are dedicated to explicit tracking such as security cameras, time-series occupancy traces, as aggregated environmental measurements, are typically deemed privacy-preserving. Unfortunately, it may still reveal some of the same sensitive information as privacy-invasive sensing such as video surveillance. We conduct experiments using a publicly available dataset and synthetic data. Our results demonstrate the underlying privacy leakage via occupancy data. We further show how our evaluation can enable adaptive privacy mechanisms to control the information leakage by the sensing system.

A review on occupancy prediction through machine learning for enhancing energy efficiency, air quality and thermal comfort in the built environment

“…An alternative way is to detect certain occupancy patterns in a particular zone rather than target individuals [65]. Also, occupancy location can be inferred from the occupancy data with some auxiliary information [66]. For instance, a purposely defocused camera that creates a 'fuzzy' or 'warped' image or out-of-focus images is also a solution to room occupancy sensing [67].…”

A review of preserving privacy in data collected from buildings with differential privacy

“…While the examples related to buildings described above mostly focus on meter data as a basis to infer, there are other measurements typically being collected in buildings. Wang and Tague [32] focused on occupancy sensor measurements that can be stored in the building automation system (BAS, or Building Management System [BMS]), which provides timeseries occupancy status (occupied/unoccupied or even counts of occupants) in each room installed with an occupancy sensor. And these data are considered valuable as an input for optimally controlling the heating, ventilation, and air conditioning (HVAC) system in a building.…”

Towards context-free semantic localisation

“…Analytics based on movements from indoor localisation are not new, but it is often based on occupancy data which can differ with different localisation systems. Wang et al [21] propose a way to infer indoor location from occupancy data and note how occupancy data may not preserve privacy. Yaeli et al [25] at IBM applied indoor localisation to understand customers behaviour by tracking shoppers and matching their data with POS data.…”