Public Transport

https://doi.org/10.1007/s12469-023-00349-0

ORIGINAL RESEARCH

A low‑cost automatic people‑counting system at bus stops

using Wi‑Fi probe requests and deep learning

Cristina Pronello1 · Deepan Anbarasan1 · Felipe Spoturno1 · Giulia Terzolo1

Accepted: 17 December 2023

© The Author(s) 2024

Abstract
Counting people is an important part of people-centric applications, and the increase
in the number of IoT devices has allowed the collection of huge amounts of data
to facilitate people counting. The present study seeks to provide a novel, low-cost,
automatic people-counting system for the use at bus stops, featuring a sniffing device
that can capture Wi-Fi probe requests, and overcoming the problem of Media Access
Control (MAC) randomization using deep learning. To make manual data collec-
tion considerably easier, a “People Counter” app was designed to collect ground
truth data in order to train the model with higher accuracy. A user-friendly, oper-
ating system-independent dashboard was created to display the most relevant met-
rics. A two-step methodological approach was followed comprising device choice
and data collection; data analysis and algorithm development. For the data analysis,
three different approaches were tested, and among these a deep-learning approach
using Convolutional Recurrent Neural Network (CRNN) with Long Short-term
Memory (LSTM) architecture produced the best results. The optimal deep learning
model predicted the number of people at the stop with a mean absolute error of ~ 1.2
persons, which can be considered a good preliminary result, considering that the
experiment was done in a very complex open environment. People-counting systems
at bus stops can support better bus scheduling, improve the boarding and alighting
time of passengers, and aid the planning of integrated multi-modal transport system
networks.

Keywords Deep learning · Automated people counting · Bus stops · Internet of

Things · Wi-Fi probe requests · Neural networks

* Cristina Pronello
[email protected]
Deepan Anbarasan
[email protected]
1
Interuniversity Department of Regional and Urban Studies and Planning, Politecnico di Torino,
Turin, Italy

13
Vol.:(0123456789)
 C. Pronello et al.

1 Introduction

Transport companies can benefit from reliable real-time people-counting systems

to estimate travel demand, improve customer satisfaction, and optimize routes
(Mehmood et al. 2019; Reichl et al. 2018). In recent years, thanks to the progress
of Information and Communication Technologies (ICT), the number of Internet of
Things (IoT) devices has soared. The increase in the number of Wi-Fi access points
and client devices has enabled the collection of vast amounts of data, making people
counting easier, since many of these devices are consumer devices (Cisco 2020).
Public Transport (PT), despite its sustainability, is still used predominantly
by students and senior citizens (Kos-Łabędowicz 2019), but an improvement in
service quality could make it more attractive to the working population; that is
to say, if it had a higher frequency, increased comfort, and better accessibility
to stops. The perception of the quality of public transportation in terms of loca-
tion, waiting times, and equipment (such as benches) is influenced significantly
by the characteristics of bus stops. To determine whether bus stops are optimally
located, properly equipped, and have a frequency of buses that match demand, it
is essential to monitor the flow of people at the stop. However, transport studies
involving people counting have largely been limited to automatic onboard pas-
senger counting (Fihn and Finndahl 2011; Mccarthy et al. 2021; Mehmood et al.
2019; Nitti et al. 2020), and little attention has been given to the importance of
bus stops in improving transport infrastructure.
Traditionally, counting people at bus stops was done manually using hand-held
counters, or via manual customer surveys once or twice a year. Such approaches
are generally expensive and time-consuming, and only provide limited snapshots
on the days that the counts/surveys are completed (Reichl et al. 2018). But with
time and the advance of machine learning and artificial intelligence techniques,
new methods have been developed for automatic data collection using treadle
mats, cameras, Wi-Fi, Bluetooth, and other sensors (Charansonney 2018; Fihn
and Finndahl 2011; Oberli et al. 2010; Rakebrandt 2015; Yang et al. 2010).
A number of Wi-Fi-based passenger counting systems have been studied in
the last few years (Choi et al. 2022; Hidayat et al. 2018; Mccarthy et al. 2021;
Nitti et al. 2020), and have been shown to have varying levels of accuracy. To the
best of our knowledge, most validation of these Wi-Fi-based counting systems
has been done either in controlled environments or only over short periods (Nitti
et al. 2020; Oransirikul et al. 2014). Some studies have been done on people-
counting systems based on Wi-Fi probe requests, but very few of these studies
have focused on counting in the vicinity of local transport infrastructure in an
open environment, like a bus stop. Also, while there have been attempts to use
machine learning and deep learning (DL) methods for automatic people-counting,
using source data based, for example, on link-blockage time in a Wi-Fi network
(Ibrahim et al. 2019), on Channel State Information (CSI) from Wi-Fi signals
(Choi et al. 2022; Liu et al. 2019), and on images from video cameras (Baumann
et al. 2022), no studies, to the best of the authors’ knowledge, have used Wi-Fi
probe request packets as the source data for a CRNN-based DL method.

13
A low‑cost automatic people‑counting system at bus stops using…

In this research we were seeking to estimate the number of people waiting at bus
stops by detecting wireless devices in the vicinity and to process the collected infor-
mation in order to determine whether DL could provide good results for this context.
The structure of this paper is as follows: Sect. 2 provides an overview of various
methods employed for automated people counting, specifically focusing on Wi-Fi
sniffing and data processing approaches. The methodology used for device detec-
tion is presented in Sect. 3, followed by the results in Sect. 4. Finally, Sect. 5 con-
cludes the paper by examining the significance of the proposed model, discussing
the study’s limitations, and suggesting future research directions.

2 Literature review

Traditionally, mat sensors have been used to count passengers by measuring their
weight when boarding the bus (Basalamah 2016). However, these systems are
expensive and require periodic maintenance, and they also have low accuracy (Bau-
mann et al. 2022). Infrared sensors or cameras have been proposed as alternatives,
but these have a number of limitations, such as blind spots, weak performance in
poor visibility (Bernini et al. 2014; Liu et al. 2019; Saponara et al. 2016; Yahiaoui
et al. 2010), and a high cost (Liu et al. 2019). Bluetooth signals have also been
explored, but they are becoming less useful as a consequence of mobile devices
automatically disabling Bluetooth functionality (Nishide and Takada 2013), their
short transmission range, and their ability to operate in “hidden” mode (Schauer
et al. 2014). Oransirikul et al. (2014) tested Bluetooth device detection as a way of
counting passengers and were able to show that this approach is less reliable than
Wi-Fi sniffing. Collaborative applications have been proposed as another option,
but they require access to mobile device functions like GPS or microphones (Carrel
et al. 2015; Gao et al. 2017).
Wi-Fi based detection has become popular owing to its lower energy usage, shorter
discovery time, and a higher likelihood of devices having an active Wi-Fi interface
(Kurkcu and Ozbay 2017; Lee et al. 2012; Schauer et al. 2014; Singh et al. 2021).
The ubiquity of smartphones, tablets, and wearables has led to the widespread
use of Wi-Fi (Cisco 2020), which allows these devices to search for wireless access
points in their immediate vicinity. This search is performed using probe requests
that contain device-specific information, including the unique Media Access Con-
trol1 (MAC) address. Even when Wi-Fi is switched off, devices continue to send
out these requests to calculate device location based on the location of known Wi-Fi
access points nearby (KODY 2018). By analysing these probe requests, it is possible
to count the number of devices in the vicinity and estimate the number of people
(Mikkelsen et al. 2016; Myrvoll et al. 2017; Reichl et al. 2018). Transport for Lon-
don has reported2 that analysing data from free Wi-Fi connections can offer valuable
insights into customer movement at underground stations.

1
https://​stand​ards.​ieee.​org/​produ​cts-​progr​ams/​regau​th/.
2
https://​tfl.​gov.​uk/​corpo​rate/​priva​cy-​and-​cooki​es/​wi-​fi-​data-​colle​ction.

13
 C. Pronello et al.

Oransirikul et al. (2014) attempted to use Wi-Fi signals to count the number
of devices at a bus stop. However, the quantity of data collected was limited, as
the experiment took place only over a short period and did not account for MAC
address randomization. Because of privacy concerns, smartphone manufacturers
have implemented a mechanism for randomizing the MAC address between probe
requests, making it difficult to uniquely identify a device (Fenske et al. 2021; Matte
2017; Purvis and Dementyev 2020; Vanhoef et al. 2016). As a result, automatic peo-
ple-counting systems based on unique MAC addresses in Wi-Fi probe requests are
becoming obsolete.
Some authors have, nevertheless, proposed ways of overcoming the problem of
address randomization. These include the iABACUS system proposed by Nitti et al.
(2020), which uses a de-randomization algorithm. Other solutions involve machine
learning models, such as those proposed by Guillen-Perez and Cano (2019) and
Uras et al. (2020), but these have not been tested in scenarios with a large number of
people.
Baumann et al. (2022) judged that the progress in DL over the past few years
has unlocked fresh opportunities for businesses, thanks to the rise in storage capac-
ity, wider availability of data, and increased computing speed. Specifically, Con-
volutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) have
shown impressive advances in performing various tasks (Liu et al. 2019). CNN
algorithms apply multiple filters over the input data to detect patterns, more infor-
mation about how CNN works can be found in Lecun et al. (1998). RNN models
make use of a loop-structure to connect different layers to process sequential data.
For more information on RNNs see Rumelhart et al. (1986). RNN models are par-
ticularly suitable for applications that involve sequence data, such as speech recogni-
tion (Graves et al. 2013) and handwriting alignment (Raue et al. 2015). However,
traditional RNNs suffer from the vanishing gradient problem when dealing with
long sequences. To overcome this, Long Short-Term Memory (LSTM) architecture
is commonly used in conjunction with RNN (DiPietro and Hager 2020; Hochreiter
and Schmidhuber 1997). A few studies have combined the benefits of both CNN and
RNN, extracting information from features and then connecting the extracted fea-
tures to produce estimations (Liang et al. 2018; Wang et al. 2019).
Various DL approaches have been tried in recent years for counting people. Ibra-
him et al. (2019) proposed an RNN-based approach called "CrossCount" that maps a
sequence of temporal link-blockage patterns to a human count, using a single Wi-Fi
link and achieving an accuracy of about 55%. The accuracy of their model decreased
as the number of people increased, which could be due to limitations in data collec-
tion and to the method of superimposition that they used to simulate multi-person
data from single-person data.
Liu et al. (2019) proposed a crowd counting method called "DeepCount" that
counts the number of people using the Channel State Information (CSI) from Wi-Fi
signals, applying CNN. Its accuracy was 86.4% in a confined environment, but the
authors did not report the performance and accuracy of their model when the num-
ber of people is greater than 5. Also, their model was designed and tested in an
indoor environment, and it is doubtful that the data they collected would be relevant
to all scenarios in an outdoor environment.

13
A low‑cost automatic people‑counting system at bus stops using…

Similarly, Choi et al. (2022) constructed machine learning and Deep Neural Net-
work (DNN) models for crowd counting using Wi-Fi Channel State Information
(CSI) in a closed environment. Although the DNN models showed a slightly better
performance, the authors ultimately preferred machine learning methods because of
the high cost of model training required to achieve the system’s best performance
with DNN. Singh et al. (2020) proposed the use of DL for time-series forecasting of
crowds over a short time horizon, based on time-stamped crowd data, but DL was
not used in their study to address MAC address randomization. Liu et al. (2022)
proposed a spatial analysis model to understand passenger flow on a Bus rapid tran-
sit (BRT) system. The authors used an approach of counting the number of unique
devices based on the probe request packets, but the paper did not account for the
phenomenon of MAC address randomization.
Although DL models have been used by a number of researchers for people
counting, these models have been limited to smaller training datasets or synthetic
data for the training phase. The information sources used in these studies were Wi-Fi
CSI and link-blockage time, rather than Wi-Fi probe requests. Moreover, the meth-
ods were assessed in a controlled environment and their performance was reported
for small crowds.
There have been a few other studies relating to people counting that address
topics that we explore, although the focus of these studies has been different. The
research by Myrvoll et al. (2017), while having some relevance to our work, did
not deal with MAC address randomization. The study by Nitti et al. (2020) centers
on de-randomization without integrating deep learning, and its contextual scope is
limited to a simulated bus environment, which differs in significant respects from
the real-world context of a bus stop. In the studies by Uras et al. (2020) and Van-
hoef et al. (2016) emphasis is placed on de-randomization and uniquely identifying
devices via fingerprinting, complemented by clustering algorithms like DBSCAN
and OPTICS (we should mention that Vanhoef et al. (2016) relates neither to people
counting nor to public transport). Although both of the studies just mentioned, like
ours, use probe request data to overcome MAC address randomization, our study
also seeks to utilize the patterns and inherent variability of different features in esti-
mating the number of passengers, and it does not attempt to identify and individu-
ate unique devices. This has the additional benefit of protecting user privacy. It is a
change in approach that led us to opt for Convolutional Neural Networks (CNN) and
Recurrent Neural Networks (RNN) rather than clustering algorithms.
Tu et al. (2021) also take a different path, encompassing network events beyond
probe request packets and involving different data collection methods. In fact, the
authors place a greater reliance on network events linked to user interactions via
a router than on probe request data. As part of this observable shift in emphasis
away from probe requests, the present study has an important contribution to make,
addressing gaps mentioned above through the application of deep learning tech-
niques to real-world Wi-Fi probe request data within the dynamic context of a bus
stop.
In the context of bus stops, where it is essential to take the capacities of stops,
together with temporal patterns, into account, as underscored by Voß et al. (2020),
there have been very few studies that have used either Wi-Fi probe request packets

13
 C. Pronello et al.

or DL methods to count passengers. Designing bus stops that balance functionality

and safety hinges on a nuanced grasp of their utilization dynamics. Furthermore,
studying bus stop crowding can provide insights into the dynamics of bus stop
usage, have useful spin-offs relating to bus stop design, and potentially even help
combat congestion by reducing dwell times of buses at stops (Tirachini 2014). A
recent study by Jee et al. (2023) has also demonstrated that Wi-Fi sensors can be
used at bus stops for estimating passenger queue lengths and waiting times, offering
a cost-effective method to assess transit demand.
The study addresses gaps in people counting research by applying deep learn-
ing, more specifically CRNN, to real-world Wi-Fi probe request data within a bus
stop environment, overcoming limitations of previous studies in relation to data size,
sources, and methodologies. The research also seeks to use these methods to esti-
mate passenger numbers, while protecting user privacy. It thus offers a singular con-
tribution in the field of transportation-oriented passenger counting.
In summary, the paper makes valuable contributions as outlined below:

• The paper contributes to the existing literature by addressing limitations in peo-

ple counting research through the use of deep learning on real-world Wi-Fi probe
request data, avoiding the use of limited or synthetic data sources.
• In contrast to previous research centered on device de-randomization and iden-
tification, the study adopts a distinct approach: utilizing inherent patterns within
diverse features to estimate passenger numbers while preserving user privacy.
• Within the context of bus stop dynamics and utilization, where Wi-Fi probe
requests and DL methods have not been extensively harnessed, the paper offers
substantial insights through the application of deep learning in a real-world set-
ting, effectively bringing theory and practice together.

3 Methodology

The design of our proposed system using wireless devices for estimating the number
of people waiting at bus stops is illustrated in Fig. 1. A device is installed at a bus
stop to detect waiting passengers, and the data is saved to a cloud database. An algo-
rithm is then applied to this data to estimate the number of people waiting for a bus,
and the output is stored in the same database.
The methodological approach comprises two steps:

• Choice of device and data collection;

• Data analysis and development of algorithms.

The proposed system also includes a dashboard for displaying the people-count
information in an easily understandable format. The dashboard was developed using
Flutter,3 an open-source user interface software development kit. The code is written

3
https://​flutt​er.​dev/.

13
A low‑cost automatic people‑counting system at bus stops using…

Fig. 1  Overview of the system

in the Dart programming language. This framework was chosen since it is highly
customizable, powerful, and cross-platform, meaning that it can be installed on a
large selection of devices and operating systems. It can be stored on a server and is
accessible from the web via standard browsers.

3.1 Device choice and data collection

The hardware board selected for Wi-Fi probe request sniffing at bus stops must have
dual Wi-Fi capabilities, be low-cost and energy-efficient, have small dimensions for
ease of installation, and possess reasonable memory capacity to collect and send
data. These requirements ensure that the device will be able to sniff and upload data,
be cost-effective for large-scale installation, occupy minimal space, and have suf-
ficient memory to store data temporarily. We identified the best options available on
the market, and these are presented in Table 1.
The cloud system for Wi-Fi probe request sniffing at bus stops must have the
capability to store sniffed packets in a database, process Hypertext Transfer Proto-
col4 (HTTP) requests for data transfer, maintain a database for the entire system,
and allow for high scalability. The device connects to the internet using Wi-Fi pro-
vided over a mobile hotspot that uses LTE. The system design means that it can
easily be extended to new stops, and transport operators can integrate data smoothly
into their systems or create a custom dashboard. The system uses a REpresentational
State Transfer5 (REST) Application Programming Interface6 (API), and a multi-plat-
form dashboard has been developed for easy access to collected and processed data.

4
https://​www.​w3sch​ools.​com/​tags/​ref_​httpm​ethods.​asp.
5
https://​restf​ulapi.​net/.
6
https://​dl.​acm.​org/​doi/​10.​1145/​800297.​811532

13
13
Table 1  Suitable marketed hardware solutions
Hardware name Capabilities Cost in Euros

Cypress pioneer KIT PSoC 6 (Infineon Technologies AG 2022) Dual Wi-Fi capability, 4 MB FRAM, and 288 KB SRAM € 96.53a
HANI IoT Board (RELOC 2022) No dual Wi-Fi capability, 320 KB of SRAM € 79.83b
Arduino UNO 33 IoT + Arduino nano 33 IoT (Arduino 2022a, b) No dual Wi-Fi capability, 520 KB of SRAM Memory € 44.80
PyTrack 2.0 X + WiPy 3.0 (Pycom 2022a, b) Dual Wi-Fi capability, 4 MB of RAM € 62.60

FRAM Ferroelectric RAM, SRAM Static RAM

a
Infineon-Cypress/CY8CKIT-062-WIFI-BT (2022)
b
Arrow Development Tools (2023)
C. Pronello et al.
A low‑cost automatic people‑counting system at bus stops using…

Scalability is critical for the quick and cost-effective installation of sniffing devices
at a large number of bus stops in a city, and to this end a plug-and-play approach is
used based on a configuration file.
Our collection of data at bus stops implied a number of choices:

(a) Choice of bus stops. These were selected using two criteria:
• The bus line needed to include bus stops at different types of locations, that
is to say crowded central urban areas in addition to less populated suburban
areas (line 55 operated by Gruppo Trasporti Torinese7 – GTT – is an example
of a line meeting this requirement);
• Stops had to be sufficiently distant from traffic intersections to avoid an exces-
sive level of “noise” in the form of Wi-Fi probe request packets from devices
in nearby stationary vehicles;
(b) Choice of the period for data collection (for automated and manual count-
ing). Data collection took place for 9 days from March 1st to April 10th 2022,
including weekends and public holidays, in three time slots: morning peak hours
(7:30–10:30), midday peak hours (12:00 to 15:00), and evening peak hours
(17:00–19:00);
(c) Choice of the characteristics of the Wi-Fi probe request packets to be used
for automated counting. Some of the features used as input for the sequential
(Sect. 3.2.2) and neural network approach (Sect. 3.2.3) for the Wi-Fi probe
request data collection are reported in Table 2;
(d) Choice of the manual counting method. The manual counting was carried out
using the “People Counter” mobile app, created for this purpose, that allows the
user to increment, decrement, and reset the number of people recorded as waiting
at the bus stop. The number of people, as soon as it changes, is immediately sent
to Firebase Firestore8 for storage. The back-end system integrates the “ground
truth” with the data taken from the sniffing device. Some screenshots from the
app are shown in Fig. 2. People waiting were counted manually in this way when
they came within 11 m of the device (although at some stops people were waiting
for a bus beyond this 11-m radius).

3.2 Data analysis and algorithm development

To analyze the data coming from the IoT sensors and estimate the number of peo-
ple waiting at the bus stop, different approaches were tested. In what follows, an
“event” will be considered as a single packet sniffed by the IoT device and posted
to the cloud database. Sniffed packets coming from Wi-Fi devices are asynchronous
events occurring on a continuous-time basis. To process the information collected
by the IoT device through synchronous discrete-time algorithms, there must be a

7
https://​www.​gtt.​to.​it/​cms/.
8
https://​fireb​ase.​google.​com/​docs/​fires​tore.

13
 Table 2  Features of Wi-Fi request packets
Feature Data type Characteristics

13
Timestamp Integer A digital record of the time of occurrence of events
MAC address A hexadecimal vector as a string Media Access Control (MAC) address is a unique identifier assigned to a Network Interface Controller
(NIC) for use as a network address in communications within a network segment
Frame aggregation Integer A feature that can mitigate the performance hit, by sending multiple data packets in a single 802.11
transmission
AMPDU Length exponent Integer Aggregated MAC Protocol Data Unit (AMPDU) improves data transmission by aggregating or grouping
several MPDU blocks. This field can take the values between 0–7 and is used to communicate the size
of the AMPDU that may be transmitted (Gast 2013)
Channel Integer Wi-Fi channels are the medium through which a wireless internet sends and receives data
FEC coding Integer Forward Error Correction (FEC) is a method to control errors in data transmission
MCS Integer The Modulation Coding Scheme (MCS) index is a metric based on several parameters of a Wi-Fi con-
nection between two stations. Namely, for 802.11ac, it depends on the modulation type, the coding rate,
the number of spatial streams, the channel width, and the guard interval
Noise floor Signed Integer The noise floor is the measure of the signal created from the sum of all the noise sources and unwanted
signals within a measurement system, where noise is defined as any signal other than the one being
monitored
RSSI Signed Integer The Received Signal Strength Indicator (RSSI) is a measurement of the power present in a received radio
signal
SGI Integer Short Guard Interval (SGI) is intended to avoid signal loss from the multipath effect
STBC Integer Space–Time Block Coding (STBC) is an optional feature of the 802.11ac standard, which allows a trans-
mitter to transfer multiple copies of data streams using an array of antennas
SSID Alphanumeric string The Service Set Identifier (SSID) is the name of the wireless network, also known as the Network Identi-
fier or ID
Sequence number Integer The sequence number is a counter used to keep track of every byte sent outward by a host
Supported rates Integer Amount of information that can be exchanged per second expressed in Megabits per second (Mbps)
C. Pronello et al.
A low‑cost automatic people‑counting system at bus stops using…

Fig. 2  People Counter app screenshots: a Search screen, b Count screen, c Plot screen

Fig. 3  Data preparation for the processing algorithms

discretization of time in the processing pipeline. The discretization that was imple-
mented groups together events occurring inside different time windows ti of length
∆t, and the algorithm takes a fixed number of these time windows N∆t as input. As
an illustration, consider the depiction in Fig. 3. Here we have four discrete time win-
dows—t1, ­t2, ­t3 and t­4—each of length ∆t. The first three events, from two distinct
devices, are assigned to time window t1. Subsequently, the fourth event is assigned
to time window t2, and so on, depending on the time of arrival of the new event.
So, for each time window, the input data will be a matrix with the number of col-
umns corresponding to the number of features from Table 2 and the number of rows

13
 C. Pronello et al.

corresponding to the number of probe request packets received in that time window.
When events are used to estimate the number of people at the bus stop, multiple
time windows are grouped together to form a “visibility period” and the data is pro-
vided in the form of multiple matrices, where the number of matrices is equal to
the number of time windows N∆t. The average stop occupancy is estimated for the
visibility period and compared with the ground truth. In order to facilitate alignment
with the ground truth data, the ground truth information is segmented into time win-
dows of equivalent duration.
At a given bus stop, each time window ti of length ∆t has a corresponding value
Pi, that is the number of people at the stop at the end of the time window. The ground
truth values are calculated as shown in Eq. (1) for a given visibility period, which
can be described as the time period during which the actual average stop occupancy
is compared with the people count calculated as described below.
NΔt
1 ∑
R̂ = P (1)
NΔt t =1 i
i

where:

• ti: Time window of length ∆t;

• Pi: number of people at the bus stop in each time window ti;
• N∆t: Number of time windows of length ∆t considered for obtaining ground truth
for a visibility period, which can be denoted in seconds as ∆t* ­N∆t;
• Ȓ: Ground truth people count for the visibility period (∆t* ­N∆t).

The number of people may be estimated from the detected packets using either
of two approaches: a packet-content independent approach, or a packet-content
dependent approach. As regards the three different methods proposed in this study,
the first method, which uses a packet-content independent approach, is presented
in Sect. 3.2.1, while the other two methods, using a packet-content dependent
approach, are presented in Sects. 3.2.2 and 3.2.3.

3.2.1 Matched digital filter

In a packet-content independent approach, the content of the detected packets is

ignored. Packets are treated simply as “events” to be counted, and then filtered as
in a classical signal-processing task. The input of these algorithms is the number of
events (sniffed packets) at each time.
The hypothesis is that the number of events observed in a given time window
(ti) is determined by the sum of two factors: the number of people (Pi) at the bus
stop and a random variable (η) that encompasses various external factors such as the
interference of nearby devices, fixed Wi-Fi antennas, and other perturbations. This
relationship is expressed by Eq. (2):

13
A low‑cost automatic people‑counting system at bus stops using…

𝜇 = Pi + 𝜂 (2)
where:

• μ corresponds to the number of events (probe requests) captured in time window

ti;
• Pi corresponds to the number of people at the bus stop at the end of the time win-
dow ti;
• η corresponds to a random variable representing noise.

Pi and η are not zero-centered signals. The first pre-processing stage serves to
center the signal μ to the value E(Pi ) that can be determined from the training data,
producing a pre-processed output ye. The idea is to take the centered signal of the
random variable as a “noise” signal to be filtered through a filter h, to produce the
estimation of the number of persons as in Eq. (3).
̂ = ye ∗ h
P (3)

where:

• h – filter to be applied to pre-processed input signal;

• ye – input signal for the matched filter (output of pre-processing);
̂ – estimated people count at a stop during time window t .
• P i

The choice of testing a matched filter is in fact to find a baseline performance that
can be obtained with a simple packet-content independent approach, and then to
check whether a packet-content dependent approach can outperform this baseline.

3.2.2 Sequential processing algorithm

The packet-content dependent approach analyses the sequence of logged packets

to estimate the number of people at a bus stop. The input to the algorithm is the
sequence of packets sniffed at each time. Before analyzing the relevant features of
the probe request packets, the input data are filtered based on the Received Signal
Strength Indicator (RSSI) contained in the sniffed packets. Only packets with an
RSSI greater than – 78 dB are considered in order to eliminate possible extraneous
sources of noise.
The sequential processing algorithm is used to analyze the most relevant features
of the probe request packets across N∆t time windows, each of length ∆t. To estimate
the number of people at the bus stop, the algorithm counts the number of unique
MAC addresses in each time window. To address this issue of MAC address rand-
omization (APPLE Inc. 2021; Fenske et al. 2021), a MAC address de-randomiza-
tion approach is used. This approach considers the supported rates and RSSI (see
Table 2) of each device within a time window. If the supported rates are unchanged

13
 C. Pronello et al.

and the RSSI is in a range of ± 4 from a previously detected MAC address within the
same time window, then the device is considered to be the same as the previously
detected device, as shown in Eq. (4).
MACia , if SRai = SRj and RSSIia − 4 ≤ RSSIj ≤ RSSIia + 4 and 1 ≤ a ≤ 𝜇
{
MACj =
MACj , otherwise
(4)
where:

• MACai: the MAC address of any detected device a in time window ti;
• ti,tj: successive time windows where j = i + 1;
• µ: number of probe request events captured in time window ti;
• RSSI ai andSRai—RSSI value and supported rates of the corresponding detected
device (see Table 2) in time window ti;
• MACj: the MAC address of a device being currently checked in time window tj;
• SRj is RSSIj: RSSI value and supported data rates at the time of detection for the
device with MAC address MACj.

The people count is then computed as the number of unique devices in each time
window ti, as shown in Eq. (5).
μ {
∑ 1, if MACa has not been so far in time window
P̂ = (5)
0, if MACa has been seen before in time window
a=1

where:

• ̂ : Estimated count of people in time window t ;

P i
• μ : number of events captured in time window ti;
• MACa: MAC address of each detected event after applying Eq. (4).

3.2.3 Convolutional and Recurrent Neural Networks (CRNN)

The input data are in the form of probe request packets with features (or columns)
as listed in Table 2. These data are further partitioned into distinct time windows, a
process illustrated in Fig. 3. The size of the input data is not fixed, but variable, as
a result of fluctuations in the number of probe requests received in different time
windows. These fluctuations are contingent upon several elements, including the
number of nearby devices, and factors dependent on device manufacturers and mod-
els. The current operational state of the device—be it idle mode, screen-on status,
or Wi-Fi connectivity—likewise contributes to these fluctuations (Pintor and Atzori
2021).
In order to fix the size of the input data, transmit it as sequential data to the adja-
cent layers, and then process the sequential data, a combination of CNN and RNN
is used. CNN is good for feature extraction due to its ability to identify spatial pat-
terns, while RNN is adept at handling sequential data such as time series, making
this combination well-suited for capturing both spatial and temporal aspects inherent

13
A low‑cost automatic people‑counting system at bus stops using…

Table 3  Convolutional layer operation algorithm

Algorithm 1: Convolutional layer operation
1. For each time window ∆t:
(a) For each sniffed packet:
- Apply each one of the filters in the convolutional network
(b) Sum the outputs of each filter for all the sensors in the time window
producing a 1 × Fc dimensional vector
2. Concatenate the output of the N∆t time windows producing a fixed-size N∆t ×
Fc matrix

Fig. 4  Convolutional input layer

in predicting people counts over time (Hasan et al. 2019). LSTM cells are used for
the RNN layer because of their capabilities in terms of long-term memory and their
better handling of the vanishing gradient problem (DiPietro and Hager 2020; Yu
et al. 2019). A CRNN algorithm (Liang et al. 2018; Wiest 2017) is used with a first
convolutional layer that fixes the input data size for the following layers by apply-
ing a set of filters to each collected packet, as described in the algorithm reported
in Table 3. The number of filters is denoted by Fc and is one of the parameters that
is optimized during the model training phase. This first convolutional operation is
shown in Fig. 4.
The convolutional filters at the CNN layer typically aid in capturing patterns
within the features found in each probe request packet. Any of these filters may be of
use in capturing conditions or patterns relating to RSSI values, SSID patterns, pat-
terns with MAC address values and so on. These conditions help in identifying how
many packets come from unique client devices belonging to passengers at the bus
stop. Then by summing the outputs, the final vector for each window is prepared and

13
 C. Pronello et al.

Fig. 5  Long Short-Term Memory (LSTM) and Rectified Linear Unit (ReLU) layer

concatenated to produce the final output matrix of fixed size N∆t × Fc, with the rows
containing data from the different time windows, and the columns representing the
filters that were applied to the data from the different time windows.
Each vector produced by the convolutional layer for each time window is then
sequentially provided as input to a recurrent layer (composed of LSTM cells) that
will learn the dynamics of how the patterns corresponding to the convolutional fil-
ters change over time, to produce a final estimation of the average stop occupancy
via a final Rectified Linear Unit9 (ReLU) neuron, as shown in Fig. 5. The final out-
put contains the model’s estimation of the number of people at the bus stop during
the visibility period.
The only parameters for this architecture are the number of filters Fc and the
number of LSTM cells Ln, that are optimized during the “Training campaign” using
stochastic gradient descent optimization (Hardt et al. 2016).

4 Results

Among the suitable devices identified (Table 1), the one selected was the WiPy 3.0
module in combination with the PyTrack 2.0 X electronic board equipped with a
microcontroller. WiPy 3.0 was used to capture Wi-Fi probe request broadcast pack-
ets. Probe requests are control frames sent by devices equipped with a Wi-Fi inter-
face. The device was programmed to emit light signals as follows:

• Blue, when in operation;

• Green, when sending an information packet to MongoDB;

9
https://​pytor​ch.​org/​docs/​stable/​gener​ated/​torch.​nn.​ReLU.​html.

13
A low‑cost automatic people‑counting system at bus stops using…

Fig. 6  Operations performed on device

Fig. 7  Cloud system overview

• Red, when sending a packet to MongoDB fails. In this case, the data collected
up to that point are saved in memory and the device is forced to restart. Once
restarted, before starting to capture other probe request packets, the device sends
everything it has previously saved in the memory.

The operations performed on the device are shown in Fig. 6. For every ten pack-
ets received, a JavaScript Object Notation (JSON) file is produced, which is sent to
the MongoDB database. This is done to reduce the number of database connection
requests the device makes.

13
 C. Pronello et al.

Fig. 8  Ferrucci bus stop. Source: Screen capture from google maps

Table 4  Schedule of the manual data collection

Start time End time Bus stop number Day of the week

2022-03-05 13:00 2022-03-05 13:00 (Artificially generated data) Saturday

2022-03-12 10:25 2022-03-12 11:24 624 Saturday
2022-03-23 07:53 2022-03-23 10:31 3287 Wednesday
2022-03-27 07:46 2022-03-27 10:34 3287 Sunday
2022-03-28 06:52 2022-03-28 09:37 3287 Monday
2022-03-29 07:39 2022-03-29 09:43 3287 Tuesday
2022-04-11 15:17 2022-04-11 16:15 39 Monday
2022-04-20 10:31 2022-04-20 12:15 3287 Wednesday
2022-04-21 15:24 2022-04-21 17:01 3287 Thursday

The device used to store and compute all the cloud functions is a Raspberry Pi 3
Model B. It was chosen since resources available for this research are limited, and
the use of faster, more expensive hardware would have resulted in over-dimension-
ing. Moreover, all the relevant software was designed to compile and work on any
other Linux-based machine, which would, if necessary, allow migration to a more
powerful machine with little effort. CherryPy is used to handle Representational
state transfer (REST) requests for all the components of the architecture. This makes
the system highly scalable since all transfers of data are handled using standard
JSON files and GET/POST10 requests (GET and POST are the two most common
HTTP transfer requests).
To store the data, MongoDB is used, which is a NoSQL11 document database. It
is open-source and used mainly for high-volume data storage (Győrödi et al. 2015).
It also provides scalability and flexibility. An overview of the cloud system is shown
in Fig. 7.

10
https://​restf​ulapi.​net/​http-​metho​ds/.
11
https://​www.​mongo​db.​com/​nosql-​expla​ined.

13
A low‑cost automatic people‑counting system at bus stops using…

Table 5  Distribution of collected ground truth values

Date Stop number Mean stop occu- Min passenger Max
pancy count passenger
count

05 March 2022 Test location 4 1 7

12 March 2022 624 4.83 0 12
23 March 2022 3287 2.97 0 9
27 March 2022 3287 3.48 0 11
28 March 2022 3287 2.55 0 7
29 March 2022 3287 3.08 0 9
11 April 2022 39 11.95 2 32
20 April 2022 3287 2.20 0 7
21 April 2022 3287 2.38 0 6

Table 6  Pre-processed features

Feature name Processing

MAC Address Split the string into a vector of six hexadecimal numbers.
Conversion from hexadecimal to integer for each element
SSID Conversion from string to integer through a hashing function

Table 7  Performance evaluation Algorithm Mean Absolute Error

of the tested algorithms during [Number of persons]
the preliminary analysis
Matched filter 1.78
Sequential algorithm 1.82
CRNN 1.25

For data storage in MongoDB, three types of collection were set up: (1) input col-
lection, storing all the raw data from the sniffing boards with the addition of a times-
tamp; (2) output collection, storing the results evaluated by the counting algorithm
together with a timestamp; and (3) device collection, including information about all
the active sniffing devices.
Data were collected, on the one hand via the device installed on the bus stop
bench, and on the other hand manually, at the Ferrucci bus stop 3287 on Corso Vit-
torio Emanuele II in Turin (Fig. 8), according to the schedule given in Table 4. On
March 5th, artificially generated data were collected, in order to provide at least one
dataset in a controlled environment. The data were collected in a room where all
Wi-Fi capable devices apart from the test devices were removed. The data for March
12th, 2022 and April 22nd, 2022 were collected, respectively, at the bus stops 624
and 39 to check that the model works in different contexts, so as to avoid model
overfitting (Gupta and Sharma 2022) to the selected stop.

13
 C. Pronello et al.

Table 8  Best parameter values Δt [s] NΔt Number of convo- Number of Mean absolute
for the neural network training lutional filters (Fc) LSTM Neu- error (MAE)
rons

30 10 60 5 1.14
60 10 60 5 1.26
60 5 60 5 1.35
60 5 10 1 1.68
30 10 30 5 1.54
30 5 60 5 1.52

Italic highlights the parameters for the best model and the BoldItalic
highlights the lowest error achieved by a model

More than 22,700 records of sniffed packets, and more than 1700 ground truth
labels were obtained from the collected data. The collected ground truth data with
the mean, minimum and maximum stop occupancy at the different bus stops is listed
in Table 5. Stop number 39 is seen to exhibit a crowding pattern that is notably dif-
ferent from the other stops in the dataset. This variance in crowding patterns pro-
vided diversity in the training and testing of the model.

4.1 Data analysis and modelling

The conversion of the input data into a processable format allows feature selection
and preparation for the analysis. Data processing algorithms need number-type data
as input, and therefore sniffed packets that may contain information in string for-
mats, such as the SSID name and the MAC address, must be converted in order to
be processed. In addition, the literature shows that for some devices, MAC addresses
are not randomized across all their fields, but only in a subset of them (Purvis and
Dementyev 2020), and so in order to let the neural network learn this process, the
MAC address was split into a vector of six numbers, as described in Table 6.
Using the collected data, the algorithms described in the methodology were
implemented and tested with different parameters iteratively, to select the param-
eters that produced the best results. The results obtained in terms of mean abso-
lute error are shown in Table 7. Based on the literature review, some studies have
suggested the use of the Mean Absolute Error (MAE) the Root Mean Square Error
(RMSE) (Brassington 2017; Willmott and Matsuura 2005) especially for datasets
with errors of varying magnitude and potential outliers. Therefore, to maintain con-
sistency among all tested methods, the MAE was chosen to be used. It can be seen
that the best-performing algorithm is the CRNN, which has the lowest MAE among
the compared algorithms. The CRNN algorithm was therefore selected for further
fine-tuning with a larger training set, varying its parameters to find the best result.
After selecting CRNN as the best algorithm for our use case, we used different
numbers of neurons for each layer, and different values for Fc, Δt and NΔt to train
different models in order to determine which performed best. Two parameters of the
model play an important role in addressing the issue of devices that come close to

13
 A low‑cost automatic people‑counting system at bus stops using…

Fig. 9  a Learning curve and b final error density distribution on the testing set

13
 C. Pronello et al.

Fig. 10  Output in time for one of the testing datasets

the bus stop but do not remain there: these are the length of the time window ∆t and
the number of time windows N∆t that form the visibility period. If the time window
duration is set too low, devices at the stop may be missed, but if it is too long there
is a greater likelihood of counting devices that are simply passing by the stop, and
there are also risks associated with MAC address randomization. The reference time
window duration values were established based on prior research (Matte 2017) and
in-house experiments in between arrival times of probe requests from client devices.
Variations from the reference value, which ranged from 30 to 60 s, were explored
in relation to fluctuations caused by device usage and model differences during the
hyper parameter optimization phase of model training. The existence of multiple time
windows helps smooth out any minor fluctuations resulting from passing devices.
The data at hand enabled us to train more than 40 models, each using distinct
parameter values. Through careful analysis, we identified the best model by select-
ing the parameter values that yielded the most favorable outcomes on the testing
dataset during the present study, as presented in Table 8. Even though more than 40
different configurations were used, Table 8 is restricted to only a few selected model
configurations, along with the configuration that produced the best result.
The learning curve on the testing set for the best model with mean absolute
error of 1.14 (Fig. 9a) shows that the algorithm improves its performance during
the training by avoiding overfitting. The final distribution for the error (computed as
error = prediction − real) is shown in Fig. 9b.
The final performance of the algorithm has a MAE of ~ 1.2 persons on the test-
ing set. The visibility period of the final algorithm (the period taken into account
in generating the people-count) is of length Δt*NΔt = 5 min, based on Δt = 30 s and
NΔt = 10, as reported in Table 8. As can be seen from the density plot for the error
(Fig. 9b), the algorithm has a positive bias, so it tends to overestimate the num-
ber of people at the bus stop. This can also be seen in Fig. 10, where the output in
time for one of the experiments is shown. The tendency to overestimate might be
due to a number of factors, such as the model requiring a larger training dataset to

13
A low‑cost automatic people‑counting system at bus stops using…

Fig. 11  Search interface (dropdown menu)

Fig. 12  Global search screen

learn underlying patterns, or practical considerations, such as passengers carrying

multiple devices, which can impact the estimation to some extent. The problem of
overestimation resulting from passengers carrying multiple devices is a more dif-
ficult problem to overcome within the scope of this study, especially if the types of
devices being carried are different—smartphones, tablets, and laptops—and if more
than one device is used around the same time.

4.2 Data visualisation through the dashboard

To facilitate the monitoring of crowding at bus stops, the dashboard has two main
screens. The first is a "search interface" (Fig. 11) featuring a dropdown menu to inform
the user of all the cities and bus stops actively in the system. Alternatively, the user
may search for and filter bus stops via city names, bus stop names, or bus numbers. As
shown in Fig. 12, the query is global, meaning that a number or a few letters can be
entered into a single text box, and the system automatically searches in all the fields.

13
 C. Pronello et al.

Fig. 13  Dashboard overview

Results are shown in real time and the user can select the entry of interest, which
then loads on the next screen. The next screen shows the people count with an
indicator of how crowded the stop is. The indicator will be one of three different
colours: green, yellow, or red. Since stops are differently dimensioned, the trans-
port operator can specify for each individual stop how numbers of people are to be
mapped to colours. On the left there is a dynamic map with a location pin, marking
the position of the bus stop. And finally, there is a plot showing the situation over
the preceding three hours. The overview screen is shown in Fig. 13. The use of Flut-
ter allows the dashboard to be adapted easily to different screen characteristics and
operating systems.

5 Discussion and conclusions

Automated people counting at bus stops has several practical benefits for public
transport management. It provides information about the number of people waiting
for buses, which can help to ensure efficient resource allocation based on demand.
Focusing on those waiting at stops, rather than on those who have actually boarded
buses, can help to better understand specific stop-related demand. It can also offer
valuable insights into passenger behaviours relating, for example, to waiting times,
seasonal fluctuations, and boarding tendencies. This kind of information is crucial
for optimizing service planning and resource distribution. An accurate count of pas-
sengers can guide informed decisions about resource allocation, even within budget
constraints. It assists in selecting appropriate locations for resource installation, con-
sidering costs and maintenance needs.
For instance, with limited budgets, decisions about installing digital displays
for real-time information or providing shelter for waiting passengers can have

13
A low‑cost automatic people‑counting system at bus stops using…

significant implications for installation and maintenance costs. These decisions can
be informed due to the knowledge of people counting systems and in understanding
crowding patterns at stops. Decisions about bus stop design can in their turn impact
bus dwell times and contribute to reducing congestion (Tirachini 2014). Overall,
automated passenger counting at bus stops contributes to more effective public
transport management and informed decision making.
To this end, the goal of this study was to estimate the number of people wait-
ing at a bus stop, with Wi-Fi probe request packets emitted by modern ICT
devices as source data, and using deep learning (DL) methodologies. The system
that was developed has a number of advantages, employing a robust, low-cost
device that is easy to install, requires little maintenance, and makes additional
bulky devices unnecessary. The optimal DL model within the current study,
using a Convolutional Recurrent Neural Network (CRNN) with Long Short-Term
Memory (LSTM) architecture for the Recurrent Neural Network (RNN), yielded
the best results, predicting the number of people waiting at the stop with a mean
absolute error of 1.2 persons. This is a promising preliminary outcome, given the
complexity of the experiment.
However, there are factors that can adversely affect the performance of the
algorithm: not everyone owns a Wi-Fi capable device; the Wi-Fi interface may
not be enabled; and there may be multiple Wi-Fi devices linked to the same
user and undetected transmissions due to mobile devices passing through areas
quickly, as also highlighted by Li et al. (2020). Additionally, environmental fac-
tors, such as noise and uncontrolled surroundings, can negatively impact the sys-
tem’s performance. In order to reduce the impact of the randomization of the
MAC addresses and background noise, and to obtain an initial model capable
of operating in a more controlled environment, the main bus stop in this study
was placed at some distance from a road intersection to provide a reference case
for future applications in more complex environments. However, other bus stops
with a more complex environment were also used in the course of this study,
including one in front of the city’s central railway station, further enriching the
breadth of the considered scenarios. However, it is important to recognize that
most of the data collection was carried out at a single location, which may limit
the generalizability of the results.
A big challenge in using the DL approach is obtaining a ground truth dataset that
is sufficiently large to train the model. The "People Counter" app was designed to
make manual counting more efficient and less resource-intensive, and to remove the
need to translate manually collected data into digital format. The reliability and effi-
ciency of this app allowed us to devote fewer resources to counting, and to increase
the quantity and the accuracy of the collected data. Previous studies on automatic
people counting evaluated their systems with limited numbers of people, such as up
to 10 people (Choi et al. 2022; Ibrahim et al. 2019), or up to 5 people in an indoor
environment (Liu et al. 2019). In contrast, the system evaluated in the present study
was tested in an open environment with up to 30 people.
Another step forward as regards previous studies, to not compromise the reli-
ability of DL models, was the extended duration of the data collection if com-
pared, for example, to Oransirikul et al. (2014), who collected data for only

13
 C. Pronello et al.

70 min, with 60 min for counting people at the bus stop. The reason is that a
limited data collection can create an imbalanced data distribution for different
count classes, which limits the availability of large amounts of training data for
deep learning models (Ibrahim et al. 2019). However, making valid comparisons
of the accuracy of our proposed people-counting system against that of existing
crowd counting systems is far from straightforward, given the different testing
approaches involved. Real data versus synthetic data is also an advantage to bal-
ance the training phase during model building, allowing an increase in accuracy,
precision, recall, and F1 as well as a lower rate of false positive and false negative
predictions (Rankin et al. 2020). The quality of synthetic data also depends on the
source data, which can introduce bias and miss random behaviors of real-world
data. Our study was able to confirm the system’s accuracy using over 14 h of
manual counting data as ground truth, with no use of synthetic data.
Given the various limitations mentioned above that are inherent in Wi-Fi sniff-
ing, a significant factor in improving our proposed system will be the amount of
data collected for training the neural network. The data must be acquired over a
long period and at different times of the day and at various types of stops. Ter-
ritorial variables such as the classification of stops based on the number of people
present and the type of territory may potentially be used as inputs to enhance the
performance of the network. Potential future research avenues include testing this
system in a variety of scenarios, including counting passengers at different types
of stops, and extending it to systems of transport other than buses, that is to say
to trams, to metros, and at train stations. Future research could explore utiliz-
ing real-time passenger counts at bus stops to refine existing models that predict
passenger arrival times, queue lengths (Jee et al. 2023) and patterns. This could
facilitate long-term planning, route scheduling, and the optimization of headways
and timetables (Olivo et al. 2019). Finally, in a future work we could consider
associating the measurements of our system with the real-time bus passing times
to monitor additional metrics.
Finally, we may assume that wireless technologies will be subject to future
developments, in the same way that MAC address randomization was introduced
a few years ago, and it is therefore important to keep abreast of new technologies
and trends with a view to proposing automated counting systems that are alterna-
tives to the well-established but more expensive camera-based systems.

Authors contributions Conceptualization: CP; Methodology: CP and DA; Formal analysis and investiga-
tion- Mobile app and Visualisation: GT; Formal analysis and investigation- Data Analysis: FS and DA;
Writing—original draft preparation: DA; Writing—review and editing: CP; Supervision: CP.

Funding Open access funding provided by Politecnico di Torino within the CRUI-CARE Agreement.

Data availability The datasets generated during and analyzed during the current study are not publicly
available due to sensitive business information that is not intended for public dissemination but are avail-
able from the corresponding author on reasonable request.

13
A low‑cost automatic people‑counting system at bus stops using…

Declarations
Conflict of interest No potential conflict of interest was reported by the authors.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License,
which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long
as you give appropriate credit to the original author(s) and the source, provide a link to the Creative
Commons licence, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended
use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permis-
sion directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/
licenses/by/4.0/.

References
APPLE inc. (2021) Wi-Fi privacy [WWW Document]. Apple Support. URL https://​suppo​r t.​apple.​
com/​guide/​secur​ity/​wi-​fi-​priva​cy-​secb9​cb314​0c/​web. Accessed 19 Aug 2022
Arduino (2022a) UNO R3 | Arduino Documentation [WWW Document]. URL https://​docs.​ardui​no.​
cc/​hardw​are/​uno-​rev3
Arduino (2022b) Nano 33 IoT | Arduino Documentation [WWW Document]. URL https://​docs.​ardui​
no.​cc/​hardw​are/​nano-​33-​iot
Arrow Development Tools (2023) HANI-IOT per Arrow Development Tools | Sistema integrato,
schede e kit di sviluppo [WWW Document]. Arrow.com. URL https://​www.​arrow.​com/​it-​it/​
produ​cts/​hani-​iot/​arrow-​devel​opment-​tools. Accessed 27 Mar 2023
Basalamah A (2016) Sensing the crowds using Bluetooth low energy tags. IEEE Access 4:4225–4233.
https://​doi.​org/​10.​1109/​ACCESS.​2016.​25942​10
Baumann D, Sommer M, Schrempp Y, Sax E (2022) Use of deep learning methods for people count-
ing in public transport. In: 2022 International Conference on Connected Vehicle and Expo
(ICCVE), pp 1–6. https://​doi.​org/​10.​1109/​ICCVE​52871.​2022.​97429​24
Bernini N, Bombini L, Buzzoni M, Cerri P, Grisleri P (2014) An embedded system for counting pas-
sengers in public transportation vehicles. In: 2014 IEEE/ASME 10th International Conference
on Mechatronic and Embedded Systems and Applications (MESA), pp 1–6. https://​doi.​org/​10.​
1109/​MESA.​2014.​69355​62
Brassington G (2017) Mean absolute error and root mean square error: which is the better metric for
assessing model performance? Presented at the EGU General Assembly Conference, p 3574
Carrel A, Lau PS, Mishalani RG, Sengupta R, Walker JL (2015) Quantifying transit travel experiences
from the users’ perspective with high-resolution smartphone and vehicle location data. Transp
Res Part C: Emerg Techn 58:224–239. https://​doi.​org/​10.​1016/j.​trc.​2015.​03.​021
Charansonney L (2018) New ways to collect traffic data, new challenges for road network authorities
(PhD thesis). Université Paris-Est
Choi H, Fujimoto M, Matsui T, Misaki S, Yasumoto K (2022) Wi-CaL: WiFi sensing and machine
learning based device-free crowd counting and localization. IEEE Access 10:24395–24410.
https://​doi.​org/​10.​1109/​ACCESS.​2022.​31558​12
Cisco (2020) Cisco Annual Internet Report - Cisco Annual Internet Report (2018–2023) White Paper
DiPietro R, Hager GD (2020) Deep learning: RNNs and LSTM. In: Zhou SK, Rueckert D, Fichtinger
G (eds) Handbook of medical image computing and computer assisted intervention, The Elsevier
and MICCAI Society Book Series. Academic Press, pp 503–519. https://​doi.​org/​10.​1016/​B978-
0-​12-​816176-​0.​00026-0
Fenske E, Brown D, Martin J, Mayberry T, Ryan P, Rye E (2021) Three years later: a study of MAC
address randomization in mobile devices and when it succeeds. Proc Priv Enhancing Technol
2021:164–181. https://​doi.​org/​10.​2478/​popets-​2021-​0042
Fihn J, Finndahl J (2011) A framework for how to make use of an automatic passenger count-
ing system. Uppsala University. Accessed from: http://​urn.​kb.​se/​resol​ve?​urn=​urn:​nbn:​se:​uu:​
diva-​158139

13
 C. Pronello et al.

Gao R, Zhao M, Ye T, Ye F, Wang Y, Luo G (2017) Smartphone-based real time vehicle tracking
in indoor parking structures. IEEE Trans Mob Comput 16:2023–2036. https://​doi.​org/​10.​1109/​
TMC.​2017.​26841​67
Gast MS (2013) 3. The MAC - 802.11ac: A Survival Guide [Book] [WWW Document]. URL https://​
www.​oreil​ly.​com/​libra​r y/​view/​80211​ac-a-​survi​val/​97814​49357​702/​ch03.​html
Graves A, Mohamed A, Hinton G (2013) Speech recognition with deep recurrent neural networks.
In: 2013 IEEE International Conference on Acoustics, Speech and Signal Processing. IEEE. pp
6645–6649. https://​doi.​org/​10.​1109/​ICASSP.​2013.​66389​47
Guillen-Perez A, Cano M-D (2019) Counting and locating people in outdoor environments: a compar-
ative experimental study using WiFi-based passive methods. ITM Web Conf 24:01010. https://​
doi.​org/​10.​1051/​itmco​nf/​20192​401010
Gupta GK, Sharma DK (2022) A review of overfitting solutions in smart depression detection mod-
els. In: 2022 9th International Conference on Computing for Sustainable Global Development
(INDIACom), pp 145–151. https://​doi.​org/​10.​23919/​INDIA​Com54​597.​2022.​97631​47
Győrödi C, Győrödi R, Pecherle G, Olah A (2015) A comparative study: MongoDB vs. MySQL. In: 2015
13th International Conference on Engineering of Modern Electric Systems (EMES), pp 1–6. https://​
doi.​org/​10.​1109/​EMES.​2015.​71584​33
Hardt M, Recht B, Singer Y (2016) Train faster, generalize better: stability of stochastic gradient descent.
In: Proceedings of the 33rd International Conference on Machine Learning, pp 1225–1234
Hasan MN, Toma RN, Nahid A-A, Islam MMM, Kim J-M (2019) Electricity theft detection in smart grid
systems: a CNN-LSTM based approach. Energies 12:3310. https://​doi.​org/​10.​3390/​en121​73310
Hidayat A, Terabe S, Yaginuma H (2018) WiFi scanner technologies for obtaining travel data about circu-
lator bus passengers: case study in obuse, Nagano Prefecture, Japan. Transp Res Rec 2672(45):45–
54. https://​doi.​org/​10.​1177/​03611​98118​776153
Hochreiter S, Schmidhuber J (1997) Long short-term memory. Neural Comput 9:1735–1780. https://​doi.​
org/​10.​1162/​neco.​1997.9.​8.​1735
Ibrahim OT, Gomaa W, Youssef M (2019) CrossCount: a deep learning system for device-free human
counting using WiFi. IEEE Sens J 19:9921–9928. https://​doi.​org/​10.​1109/​JSEN.​2019.​29285​02
Infineon Technologies AG (2022) CY8CKIT-062-WIFI-BT - Infineon Technologies [WWW Document].
URL https://​www.​infin​eon.​com/​cms/​en/​produ​ct/​evalu​ation-​boards/​cy8ck​it-​062-​wifi-​bt/
Jee H, Sun W, Schmöcker J-D, Nakamura T (2023) Demonstrating the feasibility of using Wi-Fi sen-
sors for dynamic bus-stop queue length estimation. Public Transp. https://​doi.​org/​10.​1007/​
s12469-​023-​00336-5
KODY (2018) How to track Wi-Fi devices & connect to them using probequest [WWW Document].
WonderHowTo. URL https://​null-​byte.​wonde​rhowto.​com/​how-​to/​track-​wi-​fi-​devic​es-​conne​ct-​them-​
using-​probe​quest-​01861​37/
Kos-Łabędowicz J (2019) Is there an “intergenerational gap” in transport usage? Comparing transport
preferences of seniors and students. Pr Nauk Uniw Ekon We Wrocławiu 63:58–70
Kurkcu A, Ozbay K (2017) Estimating pedestrian densities, wait times, and flows with Wi-Fi and Blue-
tooth Sensors. Transp Res Rec 2644:72–82. https://​doi.​org/​10.​3141/​2644-​09
Lecun Y, Bottou L, Bengio Y, Haffner P (1998) Gradient-based learning applied to document recogni-
tion. Proc IEEE 86:2278–2324. https://​doi.​org/​10.​1109/5.​726791
Lee S, Kim M, Kang S, Lee K, Jung I (2012) Smart scanning for mobile devices in WLANs. In: 2012
IEEE International Conference on Communications (ICC). IEEE, pp 4960–4964
Li Y, Barthelemy J, Sun S, Perez P, Moran B (2020) A case study of WiFi sniffing performance evalua-
tion. IEEE Access 8:129224–129235. https://​doi.​org/​10.​1109/​ACCESS.​2020.​30085​33
Liang K, Qin N, Huang D, Fu Y (2018) Convolutional recurrent neural network for fault diagnosis of
high-speed train bogie. Complexity 2018:e4501952. https://​doi.​org/​10.​1155/​2018/​45019​52
Liu S, Zhao Y, Xue F, Chen B, Chen X (2019) DeepCount: crowd counting with WiFi via deep learning.
https://​doi.​org/​10.​48550/​arXiv.​1903.​05316
Liu D, Zhao C, Dong H, Huang Z (2022) Spatial analysis of bus rapid transit actual operating condi-
tions: the case of Hangzhou City, China. Public Transp 14:503–519. https://​doi.​org/​10.​1007/​
s12469-​022-​00299-z
Matte C (2017) Wi-Fi tracking: fingerprinting attacks and counter-measures. Insa Lyon
Mccarthy C, Moser I, Jayaraman PP, Ghaderi H, Tan AM, Yavari A, Mehmood U, Simmons M, Weizman
Y, Georgakopoulos D, Fuss FK, Dia H (2021) A field study of internet of things-based solutions for
automatic passenger counting. IEEE Open J Intell Transp Syst 2:384–401. https://​doi.​org/​10.​1109/​
OJITS.​2021.​31110​52

13
A low‑cost automatic people‑counting system at bus stops using…

Mehmood U, Moser I, Jayaraman PP, Banerjee A (2019) Occupancy estimation using wifi: a case study
for counting passengers on busses. In: IEEE 5th World Forum Internet Things WF-IoT 2019 - Conf.
Proc. pp 165–170. https://​doi.​org/​10.​1109/​WF-​IoT.​2019.​87673​50
Mikkelsen L, Buchakchiev R, Madsen T, Schwefel HP (2016) Public transport occupancy estimation
using WLAN probing. In: Proceedings of 2016 8th International Workshop on Resilient Networks
Design and Modeling, RNDM 2016. Institute of Electrical and Electronics Engineers Inc., pp 302–
308. https://​doi.​org/​10.​1109/​RNDM.​2016.​76083​02
Myrvoll TA, Håkegård JE, Matsui T, Septier F (2017) Counting public transport passenger using WiFi
signatures of mobile devices. In: 2017 IEEE 20th Int. Conf. Intell. Transp. Syst. ITSC. https://​doi.​
org/​10.​1109/​ITSC.​2017.​83176​87
Nishide R, Takada H (2013) Detecting pedestrian flows on a mobile ad hoc network and issues with
trends and feasible applications. Int J Adv Netw Serv 6(1 & 2):108–117
Nitti M, Pinna F, Pintor L, Pilloni V, Barabino B (2020) Iabacus: A Wi-Fi-based automatic bus passenger
counting system. Energies 13:1–21. https://​doi.​org/​10.​3390/​en130​61446
Oberli C, Torres-Torriti M, Landau D (2010) Performance evaluation of UHF RFID technologies for real-
time passenger recognition in intelligent public transportation systems. IEEE Trans Intell Transp
Syst 11:748–753. https://​doi.​org/​10.​1109/​TITS.​2010.​20484​29
Olivo A, Maternini G, Barabino B (2019) Empirical study on the accuracy and precision of automatic
passenger counting in European bus services. Open Transp J 13:250–260. https://​doi.​org/​10.​2174/​
18744​47801​91301​0250
Oransirikul T, Nishide R, Piumarta I, Takada H (2014) Measuring bus passenger load by monitoring
Wi-Fi transmissions from mobile devices. Procedia Technol 18:120–125. https://​doi.​org/​10.​1016/j.​
protcy.​2014.​11.​023
Pintor L, Atzori L (2021) A dataset of labelled device Wi-Fi probe requests for MAC address de-rand-
omization. Comput Netw. Mendeley data. https://​doi.​org/​10.​17632/​j64bt​zdsdy.1
Product detail - Infineon-Cypress/CY8CKIT-062-WIFI-BT (2022) URL https://​www.​mouser.​it/​Produ​
ctDet​ail/​Infin​eon-​Cypre​ss/​CY8CK​IT-​062-​WIFI-​BT?​qs=​lYGu3​FyN48​c6Xy1​GM0bm​kg%​3D%​3D.
Accessed 20 June 2022
Purvis W, Dementyev S (2020) Get to know MAC Address Randomization in 2020. Mist. URL https://​
www.​mist.​com/​get-​to-​know-​mac-​addre​ss-​rando​mizat​ion-​in-​2020/
Pycom (2022a) Pytrack 2.0 X [WWW Document]. Pycom. URL https://​pycom.​io/​produ​ct/​pytra​ck-2-​
0-x/. Accessed 25 Sept 2022. See also https://​docs.​pycom.​io/​datas​heets/​expan​sionb​oards/​pytra​ck2/.
Accessed 07 Feb 2024
Pycom (2022b) WiPy 3.0 [WWW Document]. Pycom. URL https://​pycom.​io/​produ​ct/​wipy-3-​0/.
Accessed 25 Sept 2022. See also https://​docs.​pycom.​io/​datas​heets/​devel​opment/​wipy3/. Accessed
07 Feb 2024
Rakebrandt A (2015) Defining APC Accuracy Standards in North America [WWW Document]. Mass
Transit. URL https://​www.​masst​ransi​tmag.​com/​techn​ology/​artic​le/​12128​610/​defin​ing-​apc-​accur​
acy-​stand​ards-​in-​north-​ameri​ca. Accessed 2 July 2022
Rankin D, Black M, Bond R, Wallace J, Mulvenna M, Epelde G (2020) Reliability of supervised machine
learning using synthetic data in health care: model to preserve privacy for data sharing. JMIR Med
Inf 8:18910. https://​doi.​org/​10.​2196/​18910
Raue F, Byeon W, Breuel TM, Liwicki M (2015) Parallel sequence classification using recurrent neural
networks and alignment. In: 2015 13th International Conference on Document Analysis and Recog-
nition (ICDAR. IEEE, pp 581–585. https://​doi.​org/​10.​1109/​ICDAR.​2015.​73338​28
Reichl P, Oh B, Ravitharan R, Stafford M (2018) Using Wifi Technologies to Count Passengers in Real-
time around Rail Infrastructure. In: 2018 International Conference on Intelligent Rail Transportation
(ICIRT), pp 1–5. https://​doi.​org/​10.​1109/​ICIRT.​2018.​86415​95
RELOC (2022) HANI IoT Board. RELOC Join Internet Things Revolut. Today. URL https://​www.​reloc.​
it/​produ​cts/​hani-​iot-​board/. See also https://​www.​reloc.​it/​news-​hani-​iot-​arm-​mbed-2/. Accessed 07
Feb 2024
Rumelhart DE, Hinton GE, Williams RJ (1986) Learning representations by back-propagating errors.
Nature 323:533–536. https://​doi.​org/​10.​1038/​32353​3a0
Saponara S, Pilato L, Fanucci L (2016) Exploiting CCTV camera system for advanced passenger services
on-board trains. In: 2016 IEEE International Smart Cities Conference (ISC2), pp 1–6. https://​doi.​
org/​10.​1109/​ISC2.​2016.​75807​48
Schauer L, Werner M, Marcus P (2014) Estimating crowd densities and pedestrian flows using Wi-Fi
and Bluetooth. In: MobiQuitous 2014 - 11th International Conference on Mobile and Ubiquitous

13
 C. Pronello et al.

Systems: Computing, Networking and Services. pp 171–177. https://​doi.​org/​10.​4108/​icst.​mobiq​

uitous.​2014.​257870
Singh U, Determe J-F, Doncker P, Horlin F (2020) Crowd forecasting based on WiFi sensors and LSTM
neural networks. IEEE Trans Instrum Meas 69(9):6121–6131. https://​doi.​org/​10.​1109/​TIM.​2020.​
29695​88
Singh U, Determe J-F, Horlin F, De Doncker P (2021) Crowd monitoring: state-of-the-art and future
directions. IETE Tech Rev 38:578–594. https://​doi.​org/​10.​1080/​02564​602.​2020.​18031​52
Tirachini A (2014) The economics and engineering of bus stops: spacing, design and congestion. Transp
Res Part A Policy Pract 59:37–57. https://​doi.​org/​10.​1016/j.​tra.​2013.​10.​010
Tu L, Wang S, Zhang D, Zhang F, He T (2021) ViFi-MobiScanner: observe human mobility via vehicu-
lar internet service. IEEE Trans Intell Transp Syst 22:280–292. https://​doi.​org/​10.​1109/​TITS.​2019.​
29567​44
Uras M, Cossu R, Ferrara E, Bagdasar O, Liotta A, Atzori L (2020) WiFi probes sniffing: an artificial
intelligence based approach for MAC addresses de-randomization. In: IEEE 25th International
Workshop on Computer Aided Modeling and Design of Communication Links and Networks
(CAMAD), pp 1–6. https://​doi.​org/​10.​1109/​CAMAD​50429.​2020.​92092​57
Vanhoef M, Matte C, Cunche M, Cardoso LS, Piessens F (2016) Why MAC address randomization is
not enough: an analysis of Wi-Fi network discovery mechanisms. In: ASIA CCS 2016 - Proc. 11th
ACM Asia Conf Comput Commun Secur. pp 413–424. https://​doi.​org/​10.​1145/​28978​45.​28978​83
Voß S, Mejia G, Voß A (2020) Mystery shopping in public transport: the case of bus station design. In:
Stephanidis C, Marcus A, Rosenzweig E, Rau P-LP, Moallem A, Rauterberg M (eds) HCI interna-
tional 2020 - late breaking papers: user experience design and case studies. Lecture notes in com-
puter science. Springer, Cham, pp 527–542. https://​doi.​org/​10.​1007/​978-3-​030-​60114-0_​36
Wang R, Li Z, Cao J, Chen T, Wang L (2019) Convolutional recurrent neural networks for text classifica-
tion. In: 2019 International Joint Conference on Neural Networks (IJCNN), pp 1–6. https://​doi.​org/​
10.​1109/​IJCNN.​2019.​88524​06
Wiest L (2017) Recurrent Neural Networks - Combination of RNN and CNN. BayernCollab. URL
https://​wiki.​tum.​de/​displ​ay/​lfdv/​Recur​rent+​Neural+​Netwo​rks+-+​Combi​nation+​of+​RNN+​and+​
CNN. Accessed 1 Nov 2022
Willmott CJ, Matsuura K (2005) Advantages of the mean absolute error (MAE) over the root mean
square error (RMSE) in assessing average model performance. Clim Res 30:79–82. https://​doi.​org/​
10.​3354/​cr030​079
Yahiaoui T, Khoudour L, Meurie C (2010) Real-time passenger counting in buses using dense stereovi-
sion. J Electron Imaging 19(3):031202. https://​doi.​org/​10.​1117/1.​34559​89
Yang H, Ozbay K, Bartin B (2010) Investigating the performance of automatic counting sensors for
pedestrian traffic data collection. In: Proceedings of the 12th World Conference on Transport
Research. Lisbon, Portugal, pp 1–11
Yu Y, Si X, Hu C, Zhang J (2019) A review of recurrent neural networks: LSTM cells and network archi-
tectures. Neural Comput 31:1235–1270. https://​doi.​org/​10.​1162/​neco_a_​01199

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps
and institutional affiliations.

13