Multi-User Low Intrusive Occupancy Detection.

Azkario Pratama,Widyawan Widyawan,Marco Aiello,Alexander Lazovik

doi:10.3390/s18030796

Azkario Pratama, Widyawan Widyawan + Show 2 more

Open Access

https://doi.org/10.3390/s18030796

Copy DOI

Abstract

Smart spaces are those that are aware of their state and can act accordingly. Among the central elements of such a state is the presence of humans and their number. For a smart office building, such information can be used for saving energy and safety purposes. While acquiring presence information is crucial, using sensing techniques that are highly intrusive, such as cameras, is often not acceptable for the building occupants. In this paper, we illustrate a proposal for occupancy detection which is low intrusive; it is based on equipment typically available in modern offices such as room-level power-metering and an app running on workers’ mobile phones. For power metering, we collect the aggregated power consumption and disaggregate the load of each device. For the mobile phone, we use the Received Signal Strength (RSS) of BLE (Bluetooth Low Energy) nodes deployed around workspaces to localize the phone in a room. We test the system in our offices. The experiments show that sensor fusion of the two sensing modalities gives 87–90% accuracy, demonstrating the effectiveness of the proposed approach.

Highlights

In 2012, commercial and residential buildings accounted for 40% of the total energy consumption and were responsible for 36% of the EU total CO2 emissions [1]
We explore the occupancy inference from Bluetooth Low Energy (BLE) data obtained from other mobile phones utilizing a model trained from only one mobile phone
As this approach is considerably intrusive, we only consider this modality as a benchmark to show the best possible occupancy inference using per-device electricity consumption

Summary

Introduction

In 2012, commercial and residential buildings accounted for 40% of the total energy consumption and were responsible for 36% of the EU total CO2 emissions [1]. Commercial office buildings have the highest energy use intensity [3]. The application of occupant-driven energy control has a central role in improving the energy efficiency of some typical power consumptions in commercial office buildings, such as lighting, and heating, ventilation and air conditioning (HVAC). Lighting and HVAC consumptions can be reduced for unoccupied spaces or adjusted based on the number of occupants. Such an effort is hampered due to the insufficient fine-grained occupancy information [4]. A typical motion sensor (e.g., Passive Infrared or PIR sensor) does not support the counting and identification of peoples’ presence in a shared workspace (i.e., only detection of binary occupancy)

Objectives

Methods

Results

Discussion

Conclusion