Design of an Urban Monitoring System for Air Quality in Smart
Cities
Andrea Marini
1
, Patrizia Mariani
2
, Alberto Garinei
1,2
, Stefania Proietti
2
, Paolo Sdringola
3
,
Massimiliano Proietti
1
, Lorenzo Menculini
1
and Marcello Marconi
1,2
1
Idea-Re S.r.l., Perugia, Italy
2
Department of Sustainability Engineering, Guglielmo Marconi University, Rome, Italy
3
ENEA Italian National Agency for New Technologies, Energy and Sustainable Economic Development, Rome, Italy
Keywords: Air Quality, Urban Monitoring, LoRaWAN, Sensors, AHP, Cellular Automata, Smart City.
Abstract: Pollution is one of the main problems faced by cities nowadays, due to the increase in emissions from
anthropogenic sources resulting from economic, industrial and demographic development. High values of
pollutants, such as atmospheric particulate matter, lead to adverse effects on the environment and human
health, causing the spread of respiratory, cardiovascular and neurological problems. For instance, recent
work shows a connection between the spread of the Covid-19 pandemic and environmental pollution. In this
context, urban monitoring of pollutants can allow to evaluate and perform actions aimed at reducing
pollution in order to safeguard citizens’ health. This study proposes a method to design an urban air quality
monitoring system. It uses the AHP multi-criteria decision-making technique to define the initial positioning
of the sensors, and the cellular automata mathematical model for the following optimization, from which the
final configuration of the network is derived. In the present case study, the monitoring concerns atmospheric
particulate matter (PM10 and PM2.5) and is carried out with six sensors that constitute a LoRaWAN
network, as often used for monitoring activities in smart cities.
1 INTRODUCTION
The economic, industrial and demographic
development of the last two centuries has led to a
considerable improvement in the quality of human
life, but it has caused at the same time significant
consequences for the environment. Indeed,
anthropogenic sources such as industrial processes,
vehicular traffic and domestic heating are identified
as the main causes of pollution (Samad & Vogt,
2020). According to World Health Organization
(2006), four main air pollutants can be identified:
particulate matter (PM10, PM2.5), nitrogen dioxide
(NO2), sulfur dioxide (SO2) and ozone (O3). In the
event that the concentrations of these pollutants
reach high values, human health is likely to be
compromised with the insurgence of respiratory,
cardiovascular and neurological problems (Ghorani-
Azam et al.,2016) and the balance of ecosystems is
put at risk (De Marco et al., 2019). During the year
2016, according to the WHO, 91% of the world
population lived in places where air quality did not
meet the levels established by the guidelines; also in
the same year, air pollution caused 4.2 million
deaths worldwide. A reduction of particulate matter
from 70 to 20 micrograms per cubic metre is
estimated to reduce mortality by 15%, also lowering
the incidence of diseases (Ambient (outdoor) air
pollution, 2018). Kurt et al. (2016) studied the
effects of pollution on the respiratory system and
identified ozone and particulate matter as the main
responsible of cardiopulmonary diseases. In
particular, children have been found to be the most
sensitive to pollution-induced effects. A study
conducted on 265 children from two Indian cities
with different levels of pollution found a greater
amount of dysfunction in the respiratory tract in
children with long-term exposure to high pollution
values (De, 2020). Moreover, scientific research
showed the role of pollutants in the spread of
viruses, especially particulate matter. A more
significant presence of the Avian Influenza Virus
was identified in air samples collected during the
days of Asian dust storms, when concentrations of
PM10 and PM2.5 are higher. This showed the role
94
Marini, A., Mariani, P., Garinei, A., Proietti, S., Sdringola, P., Proietti, M., Menculini, L. and Marconi, M.
Design of an Urban Monitoring System for Air Quality in Smart Cities.
DOI: 10.5220/0010405200940101
In Proceedings of the 10th International Conference on Smart Cities and Green ICT Systems (SMARTGREENS 2021), pages 94-101
ISBN: 978-989-758-512-8
Copyright
c
2021 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
of dust storms in the long-range transport of virus
(Chen et al., 2010). Then, following the spread of
the SARS-CoV-2 virus (known as Covid-19
pandemic), numerous studies were carried out to
evaluate the role of pollution in the spread of the
disease and its consequences on the severity of the
effects caused and on mortality rates. This disease,
similar to the severe acute respiratory syndrome
(SARS) that occurred in 2002, broke out in Wuhan
(China) in December 2019 and then spread
worldwide. In Italy, the first cases of infection were
officially reported at the end of February 2020,
particularly in the northern regions. In March, a
relationship was hypothesized between air pollution
and the spread of SARS-Cov-2 infections. A
position paper (Setti, Passarini, De Gennaro, Di
Gilio et al., 2020) on this topic was published by
some experts of the Società Italiana Medicina
Ambientale (SIMA) together with researchers from
Italian universities. The authors analysed the daily
concentration of PM10 and the number of infections
by Covid-19, for each province. They found a
relationship between the exceedances of PM10 limit
values recorded in the period February 10
th
-
February 29
th
2020 and the number of COVID-19
cases updated to March 3
rd
, considering that the
infection is diagnosed with a latency time of 14
days. One month after the publication of the position
paper, SIMA claimed to have ascertained the
presence of the new coronavirus in particulate matter
from the extraction of SARS-Cov-2 RNA (Setti,
Passarini, De Gennaro, Barbieri et al., 2020). The
analysis was carried out on 34 samples of PM10
collected for three weeks (from February 21
th
to
March 13
th
2020) in industrial sites located in the
province of Bergamo. The results were confirmed on
12 samples for the three genes E, N, RdRP used as
molecular markers. European Public Health Alliance
(2020) stated that those who live in cities with high
concentrations of pollutants are more exposed to the
risks deriving from Covid-19. This hypothesis was
made on the basis of statements made by the
European Respiratory Society (ERS): people with
chronic lung and heart diseases caused by long-term
exposure to poor air quality are less able to fight
lung infections and therefore also Covid-19. To
confirm the hypothesis, results of a study conducted
in 2003 on SARS (Cui et al., 2003) were also used.
This study found that people living in regions with a
moderate air pollution index present an 84% higher
chance of death than inhabitants of regions with a
low index. Research by Wu et al. (2020a, 2020b)
showed that, in the long run, a difference of one
microgram in the average of PM2.5 is sufficient to
increase the mortality rate of Covid-19 by 11%. The
analysis compares the levels of particulate matter
recorded in 3089 American counties with deaths for
Covid-19 until June 18th 2020 and examines several
variables: population size, hospital beds, weather,
socioeconomic and behavioural conditions. A study
on Italian territory (Fattorini & Regoli, 2020)
focused on the role of chronic exposure to air
pollutants. From the analysis of NO2, PM2.5 and
PM10 values detected in Italy in the last 4 years, it
was found that Northern Italy has been constantly
exposed to high levels of atmospheric pollution and
there is a correlation between these data and the
Covid-19 cases for 71 provinces.
In order to assess the level of atmospheric
pollution and take action to ensure good air quality,
limiting the spread of Covid-19 and other diseases,
we intend to define the design of an urban
monitoring system for air quality in smart cities of a
size similar to that of the case study. The method
takes into account the main anthropogenic sources of
air pollution and it is applied in the smart city on the
basis of the specific urban characteristics of the
place under study and with the involvement of
citizen science, creating a participatory process.
2 LITERATURE REVIEW
Collecting air quality data through monitoring
networks allows to assess pollution levels and,
where appropriate, suggests actions that are to be
taken in order to avoid the adverse effects of
pollution on the environment and human health
(Kainuma et al., 1990). The chosen measurement
points must ensure the best possible
representativeness of the area's air quality and also
take into account the location of point sources such
as industrial sites (Kibble & Harrison, 2005).
Hacıoğlu et al. (2020) pinpointed the locations of
two air quality monitoring stations among potential
urban and rural sites by using two techniques:
Analytic Hierarchy Process (AHP) and Elimination
Et Choix Traduisant la Realité III (ELECTRE III).
This was done on the basis of seven criteria:
pollution levels, security, availability of electricity,
collaborations, staff support, easy access, distance.
Both methods have identified the same positions,
thus validating each other. Mofarrah et al. (2011)
divided the study area into a grid where each square
represented a possible position for the air quality
monitoring network sensor. With the criteria of air
quality, location sensitivity, cost, population
sensitivity and population density, a fuzzy matrix of
Design of an Urban Monitoring System for Air Quality in Smart Cities
95
pairwise comparisons was formed and a score was
assigned to each potential position. The optimal
positions for the sensors were identified through the
values obtained from the Fuzzy Analytical
Hierarchy Process (FAHP) plus the degree of
representativeness of the area. FAHP method was
also used to evaluate the atmospheric environmental
quality in five cities in China (Lv & Ji, 2019),
achieving better results with an index system than
the standard air pollution index.
A mathematical model that can be used to
describe and simulate environmental phenomena
varying in time and space is the cellular automaton.
Benjavanich et al. (2017) modelled and simulated
the flow of pollution with cellular automaton in an
area of 3x3 km. A variable number of sensors was
considered, and each cell was provided with updated
levels of pollution and wind action. Marín et al.
(2020) used cellular automata to simulate the spread
of air pollution considering gravity, diffusion and
wind transport as calibration factors. Lauret et al.
(2016) combined cellular automata with artificial
neural networks to evaluate the atmospheric
dispersion of methane in 2D. In particular, the neural
networks were used for making predictions and
cellular automata for space-time simulation.
3 CASE STUDY
The case study of this research is the town of Santa
Maria degli Angeli (43°03′32″N 12°34′41″E), a part
of the Municipality of Assisi (Italy) with 8470
inhabitants. It is one of the main tourist destinations
in the region, due to the presence of important
religious sites. Over the years, the area has
experienced an important urban development,
becoming equipped with all the services necessary
for residential settlement and, in addition, also with
Figure 1: View of study area.
industrial activities, favoured by the presence of the
railway line and the proximity to the highway. These
industrial activities are mainly concentrated in the
south-west area, but there is also a foundry near the
inhabited centre. Together with the road traffic,
which concentrates on the three main axes of
connection with important road arteries and with the
nearby urban centers, these activities are the main
sources of pollution for the town (Figure 1).
4 METHODOLOGY
Based on the studies in the literature and the
importance of air pollution assessment in order to
safeguard the health of citizens, we want to propose
a design method for a low-cost urban monitoring
system of air quality that can be implemented in any
small-to-medium-sized smart city. In particular, we
propose to create a LoRaWAN network, with the
location of the sensors determined through the
application of the Analytic Hierarchy Process (AHP)
multi-criteria decision-making technique between
many potential positions and optimized through the
application of the mathematical model of cellular
automata in order to ensure the best overall coverage
of the polluted area. For the case study, the
configuration of the LoRaWAN network, which
consist of six sensors, is initially established among
twelve alternatives by use of the AHP method.
These positions are corrected using cellular
automata, assigning a transition probability
determined by the level of pollution present in the
neighbourhood of the sensor. The sensors will detect
the amount of PM10 and PM2.5 which, as discussed
in the introduction, have been showed to play a key
role in the spread of viruses.
4.1 Analytic Hierarchy Process
Analytic Hierarchy Process (AHP) is a multi-criteria
decision-making technique, developed by Thomas
Lorie Saaty in the 1970s, which allows to assign
priorities to a series of decision-making alternatives
and define them on a single scale, relating
parameters that are not directly comparable, such as
qualitative and quantitative evaluations. The method
is applied in three steps: definition of a hierarchy of
the problem, comparison of judgments and
calculation of the priority vector, hierarchical
recomposition (Analytic Hierarchy Process, n.d.).
As regards the hierarchy, the final objective is
placed at the highest level, then come the various
criteria that contribute to the objective and finally
SMARTGREENS 2021 - 10th International Conference on Smart Cities and Green ICT Systems
96
the different alternatives to be evaluated. In the
second phase, in order to evaluate how much each
criterion affects the final decision, a pairwise
comparison matrix is constructed by assigning the
judgments according to the values of the
fundamental scale (Table 1). The matrix is square
and of size equal to the number of elements of the
hierarchical level being considered. For n criteria,
with 𝑖,𝑗=1,2,..𝑛, the matrix of pairwise
comparisons is:
𝐴
=𝑎

(1)
where 𝑎

indicates how much the i-th criterion
is more important than the j-th. If 𝑎

1, the
element 𝑖 is preferred to 𝑗; if 𝑎

1, the opposite is
true. In order to make consistent judgments, it must
be established that:
𝑎

=1/𝑎

for 𝑖,𝑗=1,2,..𝑛.
𝑎

=Σ

𝑎

⋅𝑎

for all 𝑖,𝑘=1,2,..𝑛.
The same type of pairwise comparison is carried out
among the alternatives referred to each criteria.
For each matrix considered, the priority vector is
obtained from the components of the main
eigenvector 𝑤 corresponding to the main eigenvalue
λ

of the matrix 𝐴:
𝐴
∙𝑤=
λ

∙𝑤
(2)
At this point, the consistency of the assessment is
verified by calculating the Consistency Index (𝐶.𝐼.):
𝐶.𝐼.=
λ

𝑛
n1
(3)
if this is less than 10% of the Random
Inconsistency (𝑅.𝐼.) value for the corresponding is
Table 1: The fundamental scale for pairwise comparison.
Intensity of
importance
Definition
1 Equal importance
2 Weak importance
3 Moderate importance
4 Moderate plus importance
5 Strong importance
6 Strong plus importance
7 Very strong importance
8 Very, very strong importance
9 Extreme importance
Table 2: Values of Random Inconsistency (𝑅.𝐼.).
𝑛 𝑅.𝐼.
𝑛 𝑅.𝐼.
1 0.00 6 1.24
2 0.00 7 1.32
3 0.58 8 1.41
4 0.90 9 1.45
5 1.12 10 1.49
number of elements 𝑛 (Table 2), the decision
acceptable (Analytic Hierarchy Process (AHP),
n.d.). Otherwise, the reasons for the inconsistency
should be analysed and the judgments reviewed in
order to reduce the inconsistency. In the last phase,
the global weights of the alternatives are defined by
applying the principle of hierarchical composition,
determining their order of importance: the local (i.e.
within a given level) weights of each alternative are
multiplied by those of the corresponding higher-
order criteria and the products thus obtained are
added together (Latora et al., 2018).
4.2 Cellular Automata
The concept of cellular automaton was introduced
by J. von Neumann in 1947 and then applied in
practice by J.H. Conway in "Game of life" in 1968.
A cellular automaton is a discrete dynamic system:
in such model space, time and properties of the
automata can only assume a finite and countable
number of states. It consists of a set of elements,
called cells, organized in a regular spatial grid and
taking on a finite number of states. The state of each
cell at a certain moment evolves according to a
given transition rule, with the updated state of a cell
depending on the previous state of the cell itself and
the states of the neighbourhood. The latter can be of
various kinds, with most common examples
including the von Neumann, Moore and Margolus
neighbourhoods ((D’Ambrosio, 2003).
5 RESULTS
The application of the AHP method has made
possible to identify the initial configuration of the
sensor network for urban monitoring of air quality.
In defining the AHP hierarchy, the final objective
was placed at the top level, i.e. the identification of
the most significant points for the monitoring
activity, then the various criteria that contribute to
the objective and therefore determine atmospheric
pollution: home heating, traffic and presence of
Design of an Urban Monitoring System for Air Quality in Smart Cities
97
industrial activities. Potential sensor positions were
located at the lowest level of the hierarchy (Figure
2). For the case study, these are twelve and were
chosen in barycentric points of each urban sector (A-
L) identified by the three main roads axes and the
roads of major importance that lead into them
(Figure 3). Generally, the composition of the
matrices, then the attribution of judgments, and the
resulting final output are determined by a single
individual or a group decision. In this case, a mixed
approach was used: a participatory process, with the
direct involvement of citizens through
questionnaires, was used to determine the hierarchy
of the criteria and a more objective method, with a
single judgment, to evaluate the different sensor
positioning alternatives.
In the distributed questionnaire it was asked to
express which is believed to be the main source of
atmospheric pollution among home heating, traffic
and the presence of industrial activities. In addition,
it was asked how much the indicated source of
pollution was more decisive than the other two,
expressing a value in the scale from 1 to 9. The
anonymous questionnaires were distributed to a
heterogeneous sample of citizens, inhabitants of the
study area, of different ages and gender. 38
questionnaires were collected, mostly from people
over the age of 60, 19 males and 19 females, who
have been living in that area for more than 10 years
and spend the whole day there. Of 38 questionnaire
replies, 25 indicated industrial activities as the main
source of pollution, 13 indicated traffic and 0 home
heating. Given the values with which they expressed
the importance of the main source of pollution
compared to the other two, the geometric mean was
calculated and approximated to the nearest integer
number in order to compose the matrix of pairwise
comparisons. In particular, it was obtained that the
presence of industrial activities has a very, very
strong importance (value 8) compared to home
heating and strong importance (value 5) compared to
traffic; instead, traffic has a very strong importance
(value 7) compared to home heating. The same
matrix is composed of the values 1 in the main
diagonal, because it concerns the pairwise
comparison of an element with itself, and of the
reciprocal values of those already indicated,
disallowing inconsistent judgments (Table 3). The
eigenvector of the matrix was calculated and the
weight of each criterion was found: 0.0544 for home
heating, 0.2331 for traffic and 0.7125 for industrial
activities. The Consistency Index (𝐶.𝐼.) is equal to
0.12 and therefore higher than 10% of the Random
Inconsistency (𝑅.𝐼.) value for three elements. Being
Figure 2: AHP hierarchy for the selection of sensor
positions for the case study.
Figure 3: Potential positions of the air quality monitoring
sensors in the study area.
a value deriving from a group decision and having
used the geometric mean, it was still considered
acceptable, without going to review the judgments.
In fact, in the case of group decisions, three
conditions must be verified: symmetry, linear
homogeneity and concordance: the use of the
geometric mean allows to respect all three and also
to have reciprocity and separability (Analytic
Hierarchy Process, n.d.). The evaluation of the
twelve alternatives for the home heating criterion
was made on the basis of the population data in each
sector, recorded in the Municipality database. A
higher population corresponds to a higher use of
home heating. Sector B has the highest number of
inhabitants while sector L has the lowest one. The
population of each sector was compared with that of
the others and the pairwise comparisons were made
objectively, assigning the values in the fundamental
scale. Regarding the traffic criterion, the analysis
was carried out considering how each sector is
enclosed by very busy roads, therefore by the
connecting axes with the nearby urban centres and
by the highway. The values associated with each
sector were compared in pairs and the matrix was
again formed using the fundamental scale of the
AHP. The evaluation of each of the twelve
alternatives with regard to the criterion of the
SMARTGREENS 2021 - 10th International Conference on Smart Cities and Green ICT Systems
98
Table 3: Matrix of pairwise comparisons of the criteria.
Home
heating
Traffic
Industrial
activities
Home heating 1 1/7 1/8
Traffic 7 1 1/5
Industrial activities 8 5 1
presence of industrial activities was made
considering the average distance of each sector from
the foundry and the industrial area to the south-west
of the town. Similarly to the two previous criteria,
the values to include in the matrix were identified in
a very objective way. The eigenvector of each
matrix was calculated and the weights of each
alternative relating to each criterion were obtained
with the subsequent normalization. All the matrices
were found to be consistent, having obtained in the
order the following Consistency Indices ( 𝐶.𝐼.):
0.1071, 0.099 and 0.1028, all less than 10% of the
Random Inconsistency (𝑅.𝐼.) value.
Finally, the last step of the AHP method was
carried out, namely the hierarchical recomposition,
adding for each of the twelve alternatives the
products of the local weights and the weights of the
relative criteria (Table 4). The six alternatives to
which correspond the highest global weights, that is,
F, G, H, J, K and L, identify the initial configuration
of the LoRaWAN network.
In this work, cellular automata are used to
establish the final positions of the air quality
monitoring sensors, optimizing the configuration
obtained with the AHP method with the aim of
maximizing the coverage of polluted areas. Firstly,
the dimensions of the grid cells to superimpose on
the study area were established. They were defined
to be 200x200 m, thus obtaining an 11x8 grid.
Twofold information was assigned to each cell: one
variable takes into account the presence or absence
of a sensor in the cell under scrutiny, and another
one is related to the level of pollution. More in
detail, the first variable was determined from the
results of the AHP method, and the second derives
from the answers of citizens to the questionnaires.
This information forms the initial state of the
cellular automaton (Figure 4). The transition rules
guiding the system’s dynamics are defined using
Moore’s neighbourhood, which is made of eight
cells plus the starting one. At a given step during the
system evolution, the configuration determines the
set of positions of the sensors in the grid. At each
iteration the sensor can move to one of the eight
surrounding cells or remain in its current position.
Table 4: Results of AHP for the localization of monitoring
sensors.
Sector Home
heating
(0.0544)
Traffic
(0.2331)
Industrial
activities
(0.7125)
Global
weights
A 0.171 0.0214 0.0141 0.0244
B 0.3174 0.0149 0.0114 0.0288
C 0.1315 0.0546 0.0434 0.0508
D 0.0364 0.0434 0.0114 0.0202
E 0.0251 0.0159 0.0411 0.0344
F 0.08 0.0346 0.129 0.1043
G 0.1034 0.2832 0.2008 0.2147
H 0.0156 0.0271 0.1515 0.1151
I 0.0482 0.0546 0.0674 0.0634
J 0.0431 0.0689 0.2008 0.1614
K 0.0156 0.1685 0.1017 0.1126
L 0.0127 0.2128 0.0275 0.0698
The displacement of each sensor is determined
stochastically according to the following procedure:
1) The coefficient 𝑘 of polluted areas
coverage is calculated for the current
location of the sensor and for the other
future possible positions, that is, the eight
cells in its neighborhood. Given a certain
cell, the 𝑘 coefficient is defined as the
weighted sum of polluted cells within the
Moore neighborhood of the cell under
consideration; the weights are chosen to
decrease exponentially with the distance
from the central cell in which the sensor is
located. The matrix of weights is therefore
the following:
0.24 0.37 0.24
0.37 1 0.37
0.24 0.37 0.24
(4)
2) A probability 𝑝
is assigned to each
possible displacement on the basis of the
calculated coefficients:
𝑝
=
𝑒
𝑒
(5)
3) The future position of the sensor is
determined by random extraction among
the nine possibilities, according to the
probabilities 𝑝
.
The new sensor configuration of is then compared
with the previous one in order to assess whether it
determines a greater overall coverage of polluted
areas. The overall coverage is computed as the sum
of the 𝑘
coefficients of all sensors, also adding
Design of an Urban Monitoring System for Air Quality in Smart Cities
99
negative penalties if pairs of sensors lie in adjacent
cells or in the same cell. The new configuration is
accepted if it results in an increase of global
coverage, otherwise it is discarded and the system
remains in the previous configuration. According to
this rule, the positions of the sensors in the case
study were changed compared to the initial state and
the configuration shown in Figure 4 was determined,
ensuring a wider coverage of the polluted area.
6 CONCLUSIONS
This study focuses on the definition of a design
method for an air quality urban monitoring system,
useful for assessing pollution levels which derive
from different sources. The method allows to
identify the most significant positions for monitoring
within the study area. Due to the nature of the
problem, which required to evaluate different
alternatives and to take into account more criteria,
we applied the AHP multi-criteria decision-making
technique. Citizens were involved in the decision-
making through questionnaires, where they were
asked to fill in the pairwise comparison matrix of the
criteria. The group decision has identified the
following scale of criteria: industrial activities
(0.7125), traffic (0.2331) and home heating
(0.0544). The twelve sensor position alternatives
were evaluated with regard to the three criteria, in an
objective way and with a single judgment,
considering the specific features of each sector:
number of inhabitants, exposure to very busy roads
and average distance from industrial activities. The
hierarchical recomposition produced the global
weights and determined the order of preference of
the alternatives. The first six sectors, namely sectors
F, G, H, J, K and L, are the one where the six
LoRaWAN sensors for urban monitoring of
atmospheric particulate matter (PM10 and PM2.5)
should be placed. However, in order to maximize
global coverage of polluted areas, an optimization of
the mentioned AHP configuration was carried out
using a cellular automaton. After defining the grid
and the type of neighbourhood, a procedure was
devised, allowing the evolution of the state
(presence/absence of sensor) of the cells based on a
transition probability determined as a function of
coverage coefficients 𝑘 of the cells in the
neighbourhood. Using this model, the positions of
the sensors that had been found with the AHP
method were corrected to achieve greater coverage
of the polluted area, thus establishing the final
configuration of the network.
Figure 4: Evolution of sensor positions from the initial
state (a) to the final configuration (b) through the cellular
automaton.
The future development of this work will deal
with a more refined optimization of the sensors
positioning, considering levels of pollution
determined not only by the replies to the
questionnaires but also by the data actually detected
by the sensors and, importantly, the epidemiological
data regarding respiratory and cardiovascular
diseases associated with long-term exposure to high
levels of pollution. Therefore, when the sensors will
be installed in the final configuration determined in
the present study, and when a significant amount of
measurements of pollutants detected by those
sensors will have been collected, the cellular
automaton will be run again. It is important to stress
that the method to define an urban air quality
monitoring system proposed in this study lends itself
to be implemented in other smart cities, with
variable numbers of sensors and the possibility of
taking into account more pollutants.
ACKNOWLEDGEMENTS
The authors would like to thank the Municipality of
Assisi for their collaboration. The study presented in
this paper is part of the PLANET project financed to
Idea-re S.r.l. by Regione Veneto (IT) POR FESR
2014-2020 Asse I Azione 1.1.1.
REFERENCES
Ambient (outdoor) air pollution. (2018, May 2). World
Health Organization. https://www.who.int/news-room/
fact-sheets/detail/ambient-(outdoor)-air-quality-and-
health
Analytic Hierarchy Process. (n.d.). Università degli Studi
di Napoli. Retrieved September 15, 2020, from
https://www.docenti.unina.it/webdocenti-be/allegati/
materiale-didattico/557402
Analytic Hierarchy Process (AHP). (n.d.). Università Ca'
Foscari Venezia. Retrieved September 15, 2020, from
http://virgo.unive.it/paolop/papers/appunti_ahp.pdf
SMARTGREENS 2021 - 10th International Conference on Smart Cities and Green ICT Systems
100
Benjavanich, S., Ursani, Z., & Corne, D. (2017,
December). Forecasting the flow of urban pollution
with cellular automata. In 2017 Sustainable Internet
and ICT for Sustainability (SustainIT) (pp. 1-6). IEEE.
10.23919/SustainIT.2017.8379801
Chen, P. S., Tsai, F. T., Lin, C. K., Yang, C. Y., Chan, C.
C., Young, C. Y., & Lee, C. H. (2010). Ambient
influenza and avian influenza virus during dust storm
days and background days. Environmental health
perspectives, 118(9), 1211-1216.
Cui, Y., Zhang, Z. F., Froines, J., Zhao, J., Wang, H., Yu,
S. Z., & Detels, R. (2003). Air pollution and case
fatality of SARS in the People's Republic of China: an
ecologic study. Environmental Health, 2(1), 1-5.
D’Ambrosio, D. (2003). Automi Cellulari nella
modellizzazione di fenomeni complessi macroscopici e
loro ottimizzazione con Algoritmi Genetici. Università
degli Studi della Calabria. https://www.mat.unical.it/
spataro/teaching/modelli/TesiPhD_Dambrosio.pdf
De Marco, A., Proietti, C., Anav, A., Ciancarella, L.,
D'Elia, I., Fares, S., Fornasier, M. F., Fusaro, L.,
Gualtieri, M., Manes, F., Marchetto, A., Mircea, M.,
Paoletti, E., Piersanti, A., Rogora, M., Salvati, L.,
Salvatori, E., Screpanti, A., Vialetto, G., Vitale, M., &
Leonardi, C. (2019). Impacts of air pollution on
human and ecosystem health, and implications for the
National Emission Ceilings Directive: Insights from
Italy. Environment international, 125, 320-333.
De, S. (2020). Long-term ambient air pollution exposure
and respiratory impedance in children: A cross-
sectional study. Respiratory Medicine, 170, 105795.
European Public Health Alliance. (2020, March 16).
Coronavirus threat greater for polluted cities. EPHA.
https://epha.org/coronavirus-threat-greater-for-
polluted-cities/
Fattorini, D., & Regoli, F. (2020). Role of the chronic air
pollution levels in the Covid-19 outbreak risk in Italy.
Environmental Pollution, 114732.
Ghorani-Azam, A., Riahi-Zanjani, B., & Balali-Mood, M.
(2016). Effects of air pollution on human health and
practical measures for prevention in Iran. Journal of
Research in Medical Sciences, 21(1), 65.
Hacıoğlu, H., Ari, A., Özkan, A., Elbir, T., Tuncel, S. G.,
Yay, O. D., & Gaga, E. O. (2016). A new approach for
site selection of air quality monitoring stations: Multi-
criteria decision-making. Aerosol and Air Quality
Research, 16(6), 1390–1402.
Kainuma, Y., Shiozawa, K., & Okamoto, S. I. (1990).
Study of the optimal allocation of ambient air
monitoring stations. Atmospheric Environment. Part
B. Urban Atmosphere, 24(3), 395-406.
Kibble, A., & Harrison, R. (2005). Point sources of air
pollution. Occupational Medicine, 55(6), 425-431.
Kurt, O. K., Zhang, J., & Pinkerton, K. E. (2016).
Pulmonary health effects of air pollution. Current
opinion in pulmonary medicine, 22(2), 138.
Latora, A., Trapani, N., & Nicosia, G. (2018, July).
Una metodologia di Multi Criteria Decision Making
a supporto dei processi di Public Management.
Rivista italiana di Public Management, 1(2).
https://www.rivistaitalianadipublicmanagement.it/una-
metodologia-di-multi-criteria-decision-making-a-
supporto-dei-processi-di-public-management/
Lauret, P., Heymes, F., Aprin, L., & Johannet, A. (2016).
Atmospheric dispersion modeling using Artificial
Neural Network based cellular automata.
Environmental Modelling & Software, 85, 56–69.
Lv, W., & Ji, S. (2019). Atmospheric environmental
quality assessment method based on analytic hierarchy
process. Discrete & Continuous Dynamical Systems -
S, 12(4–5), 941–955.
Marı
́
n, M., Rauch, V., Rojas-Molina, A., López-Cajún, C.
S., Herrera, A., & Castaño, V. M. (2000). Cellular
automata simulation of dispersion of pollutants.
Computational Materials Science, 18(2), 132–140.
Mitchell, M., Gramss, I.T., Bornholdt, S., Gross, M., &
Pellizzari, T. (2005). Computation in Cellular
Automata: A Selected Review. Non-standard
Computation.
Mofarrah, A., Husain, T., & Alharbi, B. H., (2011).
Design of Urban Air Quality Monitoring Network:
Fuzzy Based Multi-Criteria Decision Making
Approach. Air Quality Monitoring, Assessment and
Management, 11, 25-39.
Samad, A., & Vogt, U. (2020). Investigation of urban air
quality by performing mobile measurements using a
bicycle (MOBAIR). Urban Climate, 33, 100650.
Setti, L., Passarini, F., De Gennaro, G., Barbieri, P.,
Perrone, M. G., Borelli, M., Palmisani, J., Di Gilio, A.,
Torboli, V., Fontana, F., Clemente, L., Pallavicini, A.,
Ruscio, M., Piscitelli, P., & Miani, A. (2020). SARS-
Cov-2RNA Found on Particulate Matter of Bergamo
in Northern Italy: First Evidence. Environmental
Research, 109754.
Setti, L., Passarini, F., De Gennaro, G., Di Gilio, A.,
Palmisani, J., Buono, P., Fornari, G., Perrone, M. G.,
Piazzalunga, A., Barbieri, P., Rizzo, E., & Miani, A.
(2020). Position Paper Relazione circa l’effetto
dell’inquinamento da particolato atmosferico e la
diffusione di virus nella popolazione. SIMA-Società
Italiana di Medicina Ambientale; 2020.
World Health Organization. (2006). WHO Air quality
guidelines for particulate matter, ozone, nitrogen
dioxide and sulfur dioxide: global update 2005:
summary of risk assessment. https://apps.who.int/iris/
bitstream/handle/10665/69477/WHO_SDE_PHE_OE
H_06.02_eng.pdf
Wu, X., Nethery, R. C., Sabath, M. B., Braun, D., &
Dominici, F. (2020a). Air pollution and COVID-19
mortality in the United States: Strengths and
limitations of an ecological regression analysis.
Science Advances, 6(45), eabd4049.
Wu, X., Nethery, R. C., Sabath, B. M., Braun, D., &
Dominici, F. (2020b). Exposure to air pollution and
COVID-19 mortality in the United States. medRxiv.
Design of an Urban Monitoring System for Air Quality in Smart Cities
101