Accurate occupancy estimation with WiFi and bluetooth/BLE packet capture

doi:10.1016/j.comnet.2019.106876

Computer Networks

Volume 163, 9 November 2019, 106876

https://doi.org/10.1016/j.comnet.2019.106876 Get rights and content

Abstract

The capillary spread of personal devices equipped with wireless communication capabilities has enabled a series of high level services which build on capturing and processing the data packets emitted by such devices. In this paper we tackle the problem of exploiting such a methodology to perform occupancy estimation, i.e., understanding how many people are present in a specific place, an information which is valuable in many scenarios (HVAC and lighting system control, building energy optimisation, allocation and reservation of spaces, etc.). Traditional systems for occupancy estimation rely either on environmental sensors (CO2, temperature, humidity) or video cameras, which both have drawbacks: the former have generally low accuracy, while the latter require high setup and maintenance costs. In this paper we depart from such traditional approaches and propose a cheap and accurate occupancy estimation system based on the capture of both Wi-Fi and Bluetooth or Bluetooth Low Energy management frames transmitted from users’ devices. The system, implemented on low-cost hardware, leverages a supervised learning model to adapt to different spaces and transmits estimated occupancy information to a web-based dashboard. Experimental results in both indoor and outdoor uncontrolled scenarios demonstrate the validity of the proposed solution.

Introduction

Recent CISCO reports estimate that by 2021 there will be around 11.6 billions active mobile-connected devices, exceeding the forecasted global population of 7.8 billion, that is around 1.5 mobile devices per capita. Most of these devices will be smartphones or equivalent handheld personal ready devices; however, it is expected that a fraction as large as 30% will be constituted by IoT, wearable or M2M devices (smart watches, health and fitness trackers, etc.) that will communicate to the network either directly through embedded cellular connectivity or most likely through a smartphone with other wireless communication technologies such as Wi-Fi and Bluetooth or Bluetooth Low Energy (BLE).

In parallel to such a capillary spread of mobile devices, a new set of application services is emerging by properly capturing and processing the peculiar signals emitted by such devices. Indeed, both WiFi and Bluetooth/BLE are wireless communication protocols whose management frames can be easily captured with minimal hardware equipments and processed to perform user localisation and tracking, behaviour estimation, device classification and de-anonymisation, market analysis and many others [1].

In this scenario, we focus on the very specific problem of counting how many people are present in a certain area through the capture and processing of wireless frames. The problem has several implications in many fields related to building management and optimisation: as an example, in the control of indoor lighting or Heat, Ventilation and Air Conditioning (HVAC) systems, an accurate information of the number of occupants may allow energy savings ranging from 30% up to 80%, according to several studies [2], [3], [4]. Beside energy efficiency, occupancy information can be exploited to monitor the quality of indoor living and to maintain a comfortable environment for its occupants. In some scenarios, occupancy information can be also exploited to provide services to the building administrators and occupants. As an example, keeping track of the real time occupancy of spaces is a fundamental building block for automatic monitoring systems to prevent overcrowding or to trigger alerts in case of anomalies. Also, in large company settings, knowing in real-time which offices or rooms are free may be very valuable to quickly find places where to have a meeting. Similarly, knowing how many people are present in the room devoted to lunch may be useful to plan the lunch break and avoid the infamous queues at the microwave. In universities, knowing which classrooms are not occupied can be very useful for students in search of a quiet place to study or work on a project alone or in group.

Traditional systems for occupancy estimation rely on environmental sensors (CO₂, temperature, humidity) or video cameras. However, both approaches have associated pros and cons: on the one hand environmental sensors are a cheap solution although generally not very accurate. On the other hand, camera-based systems are much more precise, but they are generally costly to deploy and maintain as well as being problematic for what concerns privacy issues.

In this paper we show that the task of occupancy estimation can be performed accurately using low-cost wireless sniffers, which capture management frames transmitted by Wi-Fi and Bluetooth personal devices even without a proper connection to a network nor user data to deliver. Wi-Fi enabled devices periodically transmit Probe Requests to collect information on the network served by access points in range. As for Bluetooth devices, Inquiry Scan frames are used to discover available devices and their information. Both types of frames are transmitted without encryption and can be easily captured with cheap off-the-shelves sniffers. Moreover they contain unique device identifiers (MAC addresses), therefore making possible to estimate the number of distinct devices present around the sniffer and ultimately providing a measure of the occupancy status of a space.

In this paper, we extend the research presented in [5], in which we developed a system able to perform occupancy estimation using only Wi-Fi frames, collected by low-cost sniffers. The system presented in this paper collects Wi-Fi Probe Requests as well as Bluetooth Inquiry Scans and Bluetooth Low Energy frames and extracts a set of features capturing diverse device behaviors and signal strength levels. The input features are fed into a supervised regression model which adapts to the specific environment under consideration and outputs the occupancy estimation. Finally, such information is delivered to a central server which provides real-time monitoring of the spaces of a building.

The paper is structured as it follows: Section 2 presents an overview of the state of the art concerning occupancy detection, with reference to different methods of estimation. Section 3 presents the implementation of the system, including the hardware design of the sniffer, the learning and estimation model as well as the data visualization service. Section 4 evaluate the proposed system with experiments conducted in different spaces of a university department, and finally Section 5 concludes the paper.

Section snippets

Related work

Several works in the last few years have tackled the problem of estimating the occupancy in indoor and outdoor spaces. The existing works can be categorised based on the type of sensing technology used in three main classes: (i) environmental-based, (ii) video-based and (iii) radio-based.

Proposed system

We perform occupancy estimation leveraging a single piece of hardware capable of sniffing Wi-Fi and Bluetooth frames. For the latter radio technology we consider both the classic version and the low-energy extension (BLE). The market already offers several low-cost options for the task at hand: as an example, the $35 Raspberry PI3 model B is equipped with Broadcom’s BCM43438 chipset, capable of both 802.11ac/b/g/n and Bluetooth Classic / BLE transmissions. Similarly, several system on chips

Experiments

In order to evaluate the performance of the proposed system, we carried out measurements in different spaces of an university campus, including both indoor and outdoor environments used for recreational, working or studying activities. The details of each experimental scenario are given in the following and summarized in Table 4:

•
Indoor Space 1 (IS1): the first experiment is conducted in an indoor university laboratory. The environment has an area of about 80 sqm and is characterized by 20 fixed

Conclusion

We proposed a system for estimating the occupancy of spaces based on the capture of Wi-Fi probe requests and Bluetooth/BLE management frames. The system is built using a single dual technology sniffer able to process received packets and create different sets of features capturing different device behaviours. Such features are later fed into two supervised learning models for performing either precise occupancy estimation or occupancy level classification. We tested our prototype in

Declaration of Competing Interest

We wish to confirm that there are no known conflicts of interest associated with this publication and there has been no significant financial support for this work that could have influenced its outcome.

Edoardo Longo received his MS in Computer Science and Engineering in 2017 from Politecnico di Milano. He is currently an assistant researcher at the Advanced Network Technologies Laboratory (ANTLab) of Politecnico di Milano. His research activities are focused on Internet of Things systems for Smart Cities, with specific focus on Real Time Localisation Systems.

References (33)

A.E. Redondi et al.
Building up knowledge through passive wifi probes
Comput. Commun.
(2018)
M. Gruber et al.
CO₂ sensors for occupancy estimations: potential in building automation applications
Energy Build.
(2014)
D. Cali et al.
CO₂ Based occupancy detection algorithm: experimental analysis and validation for office and residential buildings
Build. Environ.
(2015)
L.M. Candanedo et al.
Accurate occupancy detection of an office room from light, temperature, humidity and CO₂ measurements using statistical learning models
Energy Build.
(2016)
S.H. Ryu et al.
Development of an occupancy prediction model using indoor environmental data based on machine learning techniques
Build. Environ.
(2016)
C. Jiang et al.
Indoor occupancy estimation from carbon dioxide concentration
Energy Build.
(2016)
Y. Benezeth et al.
Towards a sensor for detecting human presence and characterizing activity
Energy Build.
(2011)
I. Amin et al.
Automated people-counting by using low-resolution infrared and visual cameras
Measurement
(2008)
F. Wahl et al.
A distributed pir-based approach for estimating people count in office environments
Computational Science and Engineering (CSE), 2012 IEEE 15th International Conference on
(2012)
S. Depatla et al.
Occupancy estimation using only WiFi power measurements
IEEE J. Sel. Areas Commun.
(2015)

J. Martin et al.

A study of mac address randomization in mobile devices and when it fails

Proc. Priv. Enhanc. Technol.

(2017)

Y. Agarwal et al.

Occupancy-driven energy management for smart building automation

Proceedings of the 2nd ACM Workshop on Embedded Sensing Systems for Energy-Efficiency in Building

(2010)

V.L. Erickson et al.

Energy efficient building environment control strategies using real-time occupancy measurements

Proceedings of the First ACM Workshop on Embedded Sensing Systems for Energy-Efficiency in Buildings

(2009)

Z. Yang et al.

A multi-sensor based occupancy estimation model for supporting demand driven HVAC operations

Proceedings of the 2012 Symposium on Simulation for Architecture and Urban Design

(2012)

P. Galluzzi et al.

Occupancy estimation using low-cost wi-fi sniffers

BalkanCom 2018, Second International Balkan Conference on Communications and Networking

(2018)

L. Zimmermann et al.

Fusion of non-intrusive environmental sensors for occupancy detection in smart homes

IEEE Internet Things J.

(2017)

Cited by (41)

WiSOM: WiFi-enabled self-adaptive system for monitoring the occupancy in smart buildings
2024, Energy
There has been extensive research on building energy saving (BES), which aims to reduce energy consumption inside buildings. One of the key solutions for energy saving in buildings is to reduce energy consumption in areas that are not occupied by inhabitants. However, effective monitoring of occupants for energy-saving purposes can be challenging due to unpredictable variations in the indoor environment, such as variations in space size, furniture arrangement, the nature of occupants’ activities (e.g., varied intensities and instances), and penetration losses of walls. Unfortunately, the existing solutions for occupancy monitoring in smart buildings, such as PIR sensors, $C O_{2}$ sensors, and cameras, etc., are expensive, require excessive maintenance, and are not adaptable to the complex variations in indoor environments. This paper introduces WiSOM, for occupancy detection that utilizes the channel state information (CSI), of commodity WiFi. The method is self-adaptive and designed to handle complex variations in indoor environments. We conducted a thorough analysis of WiSOM and evaluated it under various indoor conditions, including the impact of multipath effects, the detection of different intensities and instances of activities of daily living (ADL), and the impact of wall absorption in a real-home scenario. Our evaluation demonstrated an average detection rate of 98.25% for multipath effects, 96.5% and 98.1% for different intensities and instances of ADL, and 94.4% for wall absorption. Additionally, we assessed WiSOM’s resilience to temporal variation in the CSI and achieved a false alarm rate of less than 2%. In comparison to recent baselines, WiSOM outperformed, achieving up to a 21% improvement in detection rate within real-house scenarios.
Concepts of user-centred lighting controls for office applications: A systematic literature review
2024, Building and Environment
Due to the very limited information available during building design phase, lighting automation systems tend to follow pre-defined control curves and use normative default values. However, these are unlikely to reflect the wide variation in daily working hours and individual lighting preferences. Inadequate user modelling can result in missed energy and comfort targets, as well as insufficient light doses. Given that the building sector accounts for around one third of the world's energy demand, and lighting is one of the main consumers, a higher level of user representation in lighting control systems is essential to meet climate and environmental targets. The general practical applicability of user-centred concepts usually fails due to implementation, inadequate mapping of objectives to required parameters, availability of information and mapping in control systems. This comprehensive literature review of 160 articles evaluates the potential of individual control concepts in relation to the target criteria of lighting systems and identifies the necessary technical system components for greater applicability. The focus of the review is on daylight and artificial lighting and their application in office environments. From the results obtained, key elements for better implementation of user-centred concepts in lighting control were derived. These include zoned lighting, human-in-the-loop approaches, and sensor fusion. Post-occupancy evaluation, supported by social science methods, can help to capture relevant physiological and psychological parameters. The review concludes that post-occupancy optimisation of user-centred lighting applications offers great potential for overcoming previous design limitations and subsequently reducing performance gaps.
Developing a Hidden Markov model for occupancy prediction in high-density higher education buildings
2023, Journal of Building Engineering
The number of occupants in a building as well as the occupancy patterns and profiles have a significant impact on the building energy consumption associated with heating, cooling, ventilation, and lighting systems. A range of approaches has been used in previous studies for building occupancy prediction. However, there were limitations associated with the methods used for data collection. For instance, the prediction of occupancy based on the concentration of CO₂ level adds an additional level of uncertainty into the estimation of the building occupancy level. To avoid such uncertainties, high-resolution passive infrared video camera sensors were used to capture occupancy data for an extended period of one year. The collected occupancy data have then been used to develop a Hidden Markov Model (HMM), to predict the occupancy levels in a high-density higher education case study building. The model is trained under different scenarios to find the most appropriate set of attributes associated with occupancy data that can be used to develop the data driven HMM. The occupancy attributes considered include day, week, month, and term. Moreover, the collected dataset was optimally split into 70% for training and 30% for validation using cross-validation, which yields high prediction accuracy. The results of the prediction model under different scenarios were evaluated using root mean square error and Kullback Leibler (KL) divergence.
A innovative wavelet transformation method optimization in the noise-canceling application within intelligent building occupancy detection monitoring
2023, Heliyon
The study deals with detection of the occupation of Intelligent Building (IB) using data obtained from indirect methods with Big Data Analysis within IoT. In the area of daily living activity monitoring, one of the most challenging tasks is occupancy prediction, giving us information about people's mobility in the building. This task can be done via monitoring of $C O_{2}$ as a reliable method, which has the ambition to predict the presence of the people in specific areas. In this paper, we propose a novel hybrid system, which is based on the Support Vector Machine (SVM) prediction of the $C O_{2}$ waveform with the use of sensors that measure indoor/outdoor temperature and relative humidity. For each such prediction, we also record the gold standard $C O_{2}$ signal to objectively compare and evaluate the quality of the proposed system. Unfortunately, this prediction is often linked with a presence of predicted signal activities in the form of glitches, often having an oscillating character, which inaccurately approximates the real $C O_{2}$ signals. Thus, the difference between the gold standard and the prediction results from SVM is increasing. Therefore, we employed as the second part of the proposed system a smoothing procedure based on Wavelet transformation, which has ambitions to reduce inaccuracies in predicted signal via smoothing and increase the accuracy of the whole prediction system. The whole system is completed with an optimization procedure based on the Artificial Bee Colony (ABC) algorithm, which finally classifies the wavelet's response to recommend the most suitable wavelet settings to be used for data smoothing.
Edge-based passive crowd monitoring through WiFi Beacons
2022, Computer Communications
Citation Excerpt :
The whole analysis provided here for WiFi could be replicated in a similar way for Bluetooth or Bluetooth Low Energy (BLE). Indeed, there are many examples in the literature of similar analysis conducted via Bluetooth interfaces [17,18], achieving promising results. At the outset of this project, we configured our sensors to analyze WiFi and BLE in parallel.
Tracking people’s flows has become crucial, not only for safety and security, but also for numerous practical business applications and better management of urban spaces, facilities and services. In this paper, we proposed methodologies that, exploiting IoT technology deployed at the edge of the network, allow for the analysis of people’s movement in urban environments, both outdoors and indoors. In particular, leveraging the use of WiFi probe packets sent by smart devices carried by people on the move, we first describe an implementation of our methodology using off-the-shelf hardware to count people boarding public transportation vehicles. We then present an alternate implementation using commercial WiFi scanners connected to the edge and leveraging suitably deployed virtual network functions to process the data collected by a OneM2M IoT platform, proposing also a mobility tracking procedure that can be applied to anonymized data provided by commercial WiFi scanners. Our experimental results show that the proposed approaches to people counting and mobility detection can achieve a good level of accuracy, while overall carrying a low price tag.
MARTINI: Smart meter driven estimation of HVAC schedules and energy savings based on Wi-Fi sensing and clustering
2022, Applied Energy
HVAC systems account for a significant portion of building energy use. Nighttime setback scheduling is an Energy Conservation Measure (ECM) where cooling and heating setpoints are increased and decreased respectively during unoccupied periods with the goal of obtaining energy savings. However, knowledge of a building’s real occupancy is required to maximize the success of this measure. In addition, there is the need for a scalable way to estimate energy savings potential from ECMs that is not limited by building specific parameters and experimental or simulation modeling investments. Here, we propose MARTINI, a sMARt meTer drIveN estImation of occupant-derived Heating, Ventilation and Air Conditioning (HVAC) schedules and energy savings that leverages the ubiquity of energy smart meters and Wi-Fi infrastructure in commercial buildings. We estimate the schedules by clustering Wi-Fi derived occupancy profiles and, estimate energy savings by shifting ramp-up and setback times observed in typical operational/static load profiles that are obtained by clustering smart meter energy profiles. Our case-study results with five buildings over seven months show an average of 8.1%–10.8% (summer) and 0.2%–5.9% (fall) chilled water energy savings when HVAC system operation is aligned with occupancy. We validate our method with results from Building Energy Performance Simulation (BEPS) and find that estimated average savings of MARTINI are within 0.9%–2.4% of the BEPS predictions. In the absence of occupancy information, we can still estimate potential savings from increasing ramp-up time and decreasing setback start time. In 51 academic buildings, we find savings potentials between 1%–5%.

View all citing articles on Scopus

Alessandro Redondi is currently Assistant Professor with the Dipartimento di Elettronica, Informazione e Bioingegneria of the Politecnico di Milano, Italy. He received the MS in Computer Engineering in July 2009 and the Ph.D. in Information Engineering in February 2014, both from Politecnico di Milano. From September 2012 to April 2013 Alessandro was a visiting student at the EEE Department of the University College of London (UCL). His research activities are focused on the design and optimization of IoT systems and on network data analytics.

Matteo Cesana is currently an Associate Professor with the Dipartimento di Elettronica, Informazione e Bioingegneria of the Politecnico di Milano, Italy. He received his MS degree in Telecommunications Engineering and his Ph.D. degree in Information Engineering from Politecnico di Milano in July 2000 and in September 2004, respectively. From September 2002 to March 2003 he was a visiting researcher at the Computer Science Department of the University of California in Los Angeles (UCLA). His research activities are in the field of design, optimization and performance evaluation of wireless networks with a specific focus on communication technologies for the Internet of Things and Future Generation Cellular Networks. Dr. Cesana is an Associate Editor of the Ad Hoc Networks Journal (Elsevier).

View full text

Accurate occupancy estimation with WiFi and bluetooth/BLE packet capture

Abstract

Introduction

Section snippets

Related work

Proposed system

Experiments

Conclusion

Declaration of Competing Interest

Comput. Commun.

Energy Build.

Build. Environ.

Energy Build.

Build. Environ.

Energy Build.

Energy Build.

Measurement

IEEE J. Sel. Areas Commun.

Proc. Priv. Enhanc. Technol.

Occupancy-driven energy management for smart building automation

Proceedings of the 2nd ACM Workshop on Embedded Sensing Systems for Energy-Efficiency in Building

Energy efficient building environment control strategies using real-time occupancy measurements

Proceedings of the First ACM Workshop on Embedded Sensing Systems for Energy-Efficiency in Buildings

A multi-sensor based occupancy estimation model for supporting demand driven HVAC operations

Proceedings of the 2012 Symposium on Simulation for Architecture and Urban Design

Occupancy estimation using low-cost wi-fi sniffers

BalkanCom 2018, Second International Balkan Conference on Communications and Networking

Fusion of non-intrusive environmental sensors for occupancy detection in smart homes

IEEE Internet Things J.