ARTIFICIAL INTELLIGENCE

An opportunity for Smart cities

..

Anikeit Bagaria Arnay Verma

219301055 219301203
Sec – 5I
CSE
CS3101 Artificial Intelligence and Soft Computing
 Index
Sl.No. Title Pages No.
01 Introduction Page 1
02 Index 2
03 Abstract and Literature Review 3-19
04 Methodology 20-34
05 Implementation of Work 35-40
06 Result and Conclusion 41-45
07 References 46-50
Abstract standard of living, and facilitating wider
In the era of the Internet of things IoT access to and use of cutting-edge
and Artificial knowledge (AI) continues Information and Communication
to define the century. Artificial Technologies (ICT). In smart cities, ICT
intelligence refers to a wide term that is essential for policymaking, decision
incorporates machine learning, normal making, plan execution, and the delivery
language handling, rule based expert of useful services. The primary goal of
systems, actual robots, and robotic this analysis is to investigate the part
automation . The rise of computerized that AI and machine learning play.
system and clinical gadgets in securely Examples of ed tech include Deep
and productively diagnosing, treating, Reinforcement Learning (DRL) and
and planning care is an obvious sign Machine Learning (ML). In a
that AI is digging in for the long haul complicated smart city setting, the
and fill in significance. While AI holds a aforementioned methods can be used to
great deal of potential, the ramifications develop the best possible rules. Smart
for essential consideration suppliers transportation, cyber-security, energy
should be tended to as it might restrict efficient usage of smart grids (SG),
execution. Since the Epidemic cities in efficient use of Unmanned Aerial
2019, the healthcare industry has Vehicles (UAVs) to guarantee the best
escalated its AI-based advances by 60%. 5G and beyond 5G (B5G)
As indicated by the investigation, AI communications, and a smart health
calculations like ANN, RNN/LSTM, monitoring are all discussed in length in
CNN/R-CNN, DNN, and SVM/LS-SVMb this paper. Finally, we discuss the
has a higher impact on the different research challenges that have yet to be
smart city domains. Smart city advances met and potential future research
the unification of conventional urban directions that could bring the concept
infrastructure and information of a "smart city" closer to fruition.
technology (IT) to improve the quality of Artificial Intelligence (AI) is one of the
living and sustainable urban services in disruptive technologies that is shaping
the city. To achieve this, smart cities the future. It has growing applications
require coordinated effort among the for data-driven decisions in major smart
general public as well as private sectors city solutions, including transportation,
to introduce IT stages to gather and education, healthcare, public
examine massive amounts of governance, and power systems. At the
information. Simultaneously, it is vital same time, it is gaining popularity in
to design effective artificial intelligence protecting critical cyber infrastructure
(AI) based tools to deal with medical from cyber threats, attacks, damages, or
services emergency circumstances in unauthorized access. However, one of
smart urban communities. This paper the significant issues of those traditional
reviews about the current technologies AI technologies (e.g., deep learning) is
like artificial intelligence in the that the rapid progress in complexity
healthcare for smart cities. and sophistication propelled and turned
The goals of "smart cities" include out to be uninterpretable black boxes.
relieving the burden of growing On many occasions, it is very
urbanisation, cutting down on energy challenging to understand the decision
use, protecting the environment, and bias to control and trust systems‘
boosting the local economy and people's unexpected or seemingly unpredictable
outputs. It is acknowledged that the loss emerging to design and manage
of control over interpretability of futuristic smart cities. In this review
decision-making becomes a critical issue article, we provide an overview of smart
for many data-driven automated cities, defining their characteristics and
applications. But how may it affect the exploring the architecture of IoT. A
system‘s security and trustworthiness? detailed analysis of various wireless
This chapter conducts a comprehensive communication technologies employed
study of machine learning applications in smart city applications is presented,
in cybersecurity to indicate the need for with extensive research conducted to
explainability to address this question. determine the most appropriate
While doing that, this chapter first communication technologies for specific
discusses the black-box problems of AI use cases. The article also sheds light on
technologies for Cybersecurity different AI algorithms and their
applications in smart city-based suitability for smart city applications.
solutions. Later, considering the new Furthermore, the integration of IoT and
technological paradigm, Explainable AI in smart city scenarios is discussed,
Artificial Intelligence (XAI), this chapter emphasizing the potential contributions
discusses the transition from black-box of 5G networks coupled with AI in
to white-box. This chapter also discusses advancing modern urban environments.
the transition requirements concerning This article contributes to the existing
the interpretability, transparency, literature by highlighting the
understandability, and Explainability of tremendous opportunities presented by
AI-based technologies in applying integrating IoT and AI, paving the way
different autonomous systems in smart for the development of smart cities that
cities. Finally, it has presented some significantly enhance the quality of life
commercial XAI platforms that offer for urban dwellers while promoting
explainability over traditional AI sustainability and productivity. By
technologies before presenting future exploring the potential of IoT, AI, and
challenges and opportunities. their integration, this review article
As the global population grows, and provides valuable insights into the
urbanization becomes more prevalent, future of smart cities, demonstrating
cities often struggle to provide how these technologies can positively
convenient, secure, and sustainable impact urban environments and the
lifestyles due to the lack of necessary well-being of their inhabitants.
smart technologies. Fortunately, the Cities are increasingly turning towards
Internet of Things (IoT) has emerged as specialized technologies to address
a solution to this challenge by issues related to society, ecology,
connecting physical objects using morphology and many others. The
electronics, sensors, software, and emerging concept of Smart Cities highly
communication networks. This has encourages this prospect by promoting
transformed smart city infrastructures, the incorporation of sensors and Big
introducing various technologies that Data through the Internet of
enhance sustainability, productivity, and Things (IoT). This surge of data brings
comfort for urban dwellers. By new possibilities in the design and
leveraging Artificial Intelligence (AI) to management of cities just as much as
analyze the vast amount of IoT data economic prospects. While Big Data
available, new opportunities are processing through Artificial
Intelligence (AI) can greatly contribute that make urban life more sustainable.
to the urban fabric, sustainability and So far, the idea of using IoT-based
liveability dimensions however must not sensor networks for health care
be overlooked in favour of technological applications is a promising one with the
ones. This paper reviews the urban potential of minimizing inefficiencies in
potential of AI and proposes a new the existing infrastructure. A machine
framework binding AI technology and learning approach is key to successful
cities while ensuring the integration of implementation of the IoT-powered
key dimensions of Culture, Metabolism wireless sensor networks for this
and Governance; which are known to be purpose since there is large amount of
primordial in the successful integration data to be handled intelligently.
of Smart Cities for the compliance to Throughout this paper, it will be
the Sustainable Development Goal 11 discussed in detail how AI-powered IoT
and the New Urban Agenda. This paper and WSNs are applied in the healthcare
is aimed towards Policy Makers, Data sector. This research will be a baseline
Scientists and Engineers who are study for understanding the role of the
looking at enhancing the integration of IoT in smart cities, in particular in the
Artificial Intelligence and Big Data in health care sector, for future research
Smart Cities with an aim to increase the works.
liveability of the urban fabric while The increase in urban population
boosting economic growth and has brought climate, technological
opportunities. and economic changes that may
People want cities efficient,
negatively affect the quality of life in
technologically more advanced, greener
and more socially inclusive. These cities. In response, the concept of a
concepts include technical, economic smart city has emerged referring to
and social innovations. This term has use of novel ICTs to reduce the
been tossed around by various actors in adverse effects on cities and its
politics, business, administration and inhabitants. Among other
urban planning since the 2000s to technologies, Artificial Intelligence
establish tech-based changes and (AI) is used in that context, evolving
innovations in urban areas. The idea of
rapidly and playing an essential role
the smart city is used in conjunction
with the utilization of digital in supporting intelligent city-wide
technologies and at the same time systems in different domains. It is
represents are action to the economic, thus beneficial to identify current
social and political challenges that research advances and get a better
postindustrial understanding of the role the AI
societies are confronted with plays in this particular context.
at the start of the new millennium. The Consequently, there is a need to
key focus is on dealing with challenges
systematically study the connection
faced by urban society , such as
environmental pollution, demographic between AI and smart cities, by
change, population growth, healthcare, focusing on the findings that
the financial crisis or scarcity of uncover its role, possible
resources. In a broader sense, the term applications, but also challenges to
also includes non-technical innovations using the concepts and technologies
branded as AI in smart cities. delivered by the smart cities.
Therefore, the paper presents a Smart cities are an endeavour to
systematic literature review and make cities more efficient,
provides insights into the sustainable and liveable. In other
achievements and advances of AI in words, a smart city is a city that
smart cities pertaining to the can monitor and integrate
mentioned aspects. functionality of all the critical
The emergence of artificial infrastructure like roads, tunnels,
intelligence (AI) and its airways, waterways, railways,
progressively wider impact on many communication power supply, etc.,
sectors requires an assessment of its control maintenance
effect on the achievement of the
activities and can help in
Sustainable Development Goals.
optimizing the resources while
Using a consensus-based expert
keeping an eye on the security
elicitation process, we find that AI
can enable the accomplishment of
issues as well.
134 targets across all the goals, but This research paper explores
it may also inhibit 59 targets. various aspects and dimensions of
However, current research foci a smart city. To bridge the gap in
overlook important aspects. The literature regarding the concept
fast development of AI needs to be of smart cities and its
supported by the necessary implementation, a framework has
regulatory insight and oversight for been developed to get better
AI-based technologies to enable insights about the idea of smart
sustainable development. Failure to city. On the
do so could result in gaps in basis of extensive and deep
transparency, safety, and ethical research of literature from diverse
standards. domains, we have identified six
With the advent of ―smartphones‖, significant pillars for developing
technology has helped mankind to the framework as: Social,
solve some of its difficulties. On a Management, Economic, Legal,
similar note, ―smart Technology and Sustainability
city‖ is a futuristic approach to (SMELTS). The paper throws light
alleviate obstacles triggered by upon how these factors can make
ever-increasing population and the smart city initiative a
fast urbanization which is going to successful project. The proposed
benefit the governments as well as framework has been used to figure
the masses. Modern day cities are out various agendas for research
deprived of vital elements like and traces its practical
quality of life and socioeconomic implications.
development which can be
Smart cities have been a popular learning; sensor networks;
topic for the city stakeholders. A artificial intelligence; healthcare
smart city is the next urban Sustainability; Smart-Governance;
lifestyle that citizens expect. Due Internet of Things; Conceptual
to the hypercompetitive framework smart city; fuzzy
and globalized economy, many logic; machine learning; prediction
cities have already started or are
about to start their smart city
projects. There is
no uniform benchmark to evaluate
the
smart cities‘ performance. Several
organizations use their own
indicators
to evaluate smart cities worldwide
or
nationwide. This research paper
leverages fuzzy logic to label smart
city leaders and followers based on
various organization‘s evaluation
meta results and then uses
machine
learning techniques to identify the
key
characteristics of leaders and
followers. Based on the training
data
performance, the Support Vector
Machine (SVM) is used to predict
who will be the next smart city
leader
or follower. According to the
proposed prediction framework,
we
have successfully predicted 30
smart
city leaders and 20 followers.
Keywords: smart city,
communication technology, IoT,
5G, AI smart cities; IoT; machine
2 LITERATURE REVIEW real-time data; providing a deeper
Numerous researchers support understanding of how cities
that the processing and evolve, adapt and respond to
interpretation of data is an various conditions. Allam and
essential step towards enriching Newman (2018a) [5] support that
the urban fabric [1]. Data can now data analysis can allow cities to
be sourced from numerous support socioeconomic
neighborhoods to gain a better dimensions while ensuring the
holistic understanding of the implementation of sustainability
urban fabric. This allows planners features; including livability
and policymakers to shift from components. The processing of
closed systems; where different data through AI can ensure the
urban elements such as land plots, better provision of livability
open spaces, physical buildings dimensions; through cleanliness,
and streets are inter-linked to health and conducive
what Levy (1999) [2] calls ‗open environments for people to live
fragmented peri-urban fabric‘. and work without the urban
These have tangible impacts on challenges of pollution and
different aspects of the urban congestion. It is further believed
fabric including density, that, through this technology, the
fragmentation, cohesion, built environment can digitally
compactness and other variances support intelligent and responsive
as expressed by Lietal (2016). [3] services both conveniently and in
A better management of the real time. In addition, cities are
various dimensions and seen to leverage Big Data from AI
components of the urban fabric is to attract higher economic returns
thus crucial. This is becoming since their enabling
increasingly possible as cities infrastructures such as
around the world are becoming connectivity, energy and
digitized through the installation computing capabilities, amongst
of sensors, computational cores others, allow them to support
and different tele-communication globally competitive jobs [6]
systems (Alvalez, 2017) [4], as However Big Data can add to the
promoted by the contemporary complexity of data reliance [7].
concept of Smart Cities (Allam & For Bari (2017) the availability of
Newman, 2018a). [5] These digital Big Data poses challenges such as
concepts are tied to Artificial scaling, spanning, preparation,
Intelligence and machine learning analysis and storage bottleneck.
technologies, which are making it Fan et al. (2014) contends that Big
possible for the collection of near Data call for large sample size;
hence, the problems of noise 'urban planning', 'transportation',
accumulation, spurious 'energy efficiency', 'public safety',
correlation, measurement errors and 'citizen engagement'. We
and incidental endogeneity, which included articles that focused on
can delay or impact results. On the the impact of AI on smart cities
other hand, there are advanced and excluded those that were not
analytical trends that supports Big peer-reviewed or were not written
Data discovery; hence, quality in English. Data extraction and
analysis and interpretation. Even analysis We extracted data from
though that there are notable each selected article on the impact
challenges, the primary advantage of AI on smart cities, including the
of AI in Big Data analysis is that it specific applications of AI in urban
supports the heterogeneity and planning, transportation, energy
commonality principles that are efficiency, public safety, and
part of the Big Data analysis. citizen engagement. We organized
Through the urban fabric, AI could this information into tables and
capture hidden structures of urban synthesized the findings to identify
cells and components and provide common themes and patterns. We
a deeper understanding of used Excel and NVivo software to
common features. Some of these assist with data analysis. Quality
hidden structures are warranted assessment. We assessed the
by dynamics and complexities that quality of the selected articles
modern cities are facing due to using the PRISMA (Preferred
application of technologies in most Reporting Items for Systematic
of their fabrics.[8] The adoption of Reviews and Meta-Analyses)
AI, as the source of Big Data checklist. We evaluated each
analysis, in cities is encouraged article based on the following
since it allows for cognitive criteria: relevance to the research
computing integration; hence, aid question, study design, sample
in urban design, planning size, data collection methods, data
processes and governance. analysis, and reporting of results.
We conducted a systematic Two independent reviewers
literature review of articles assessed the quality of each article,
published between 2019 and 2023 and disagreements were resolved
in the following databases: IEEE through discussion. Limitations
Xplore, ACM Digital Library, Our study is limited by the scope
ScienceDirect, and Google Scholar. of our research question, which
We used a combination of the focuses specifically on the impact
following search terms: 'smart of AI on smart cities. We may have
cities', 'artificial intelligence', missed relevant articles that did
not use the exact search terms we and fruitful. Utilize AI, ML, and
used or that were published DRL for the most accurate analysis
outside of our date range. of enormous data sets [10, 15]. These
Additionally, our study is limited techniques have the ability to
by the quality of the selected provide near-optimal control
articles, which varied in their decisions [16] by taking a long-term
study design, sample size, and objective into account. Increasing
data collection methods the overall amount of data utilised
According to projections [9, 10], for training can aid in improving
cities will house 66 to 70 percent the accuracy and precision of the
of the global population by 2050. just described approaches [17]. The
Current urbanisation trends will authors of [18] give evidence that
impact the ecology, better data analysis for big data
administration, and security of and smart city development
cities [11]. "Smart cities" are a occurred simultaneously. As
solution for many nations to better technologies like smart cities, As
manage resources and energy and the IoT [19], blockchain technology,
keep pace with urbanisation. More than 50% of global
Smart city projects can design and population is now urbanized
implement technologies with (United Nations Report 2012) [20].
reduced carbon emissions. Several Complexity of social
nations (the United States, the ecosystem in cities and urban
European Union, and Japan) are areas has increased making
creating and implementing smart sustainability an important factor.
city plans to address problems. The ecosystem as a
Information and communication whole is experiencing an economic
technologies (ICTs) [12, 13] must be turmoil, high rate of urbanization,
utilised effectively in order to climatic variations and vast
successfully manage data population growth
processing, data transmission, and (Cisco Report 2013) [21]. These
the execution of complex factors hamper the growth and
strategies in order to guarantee a cities become disordered and
smooth and risk-free operation. unorganized (Johnson 2008) [22].
The great majority of applications Problems related to health, traffic,
for data-intensive smart cities [14] pollution, scarcity of resources,
rely heavily on the Internet of waste management and poor
Things. When dealing with large infrastructure arise and hence
and intricate information, it can be development of city falls apart.
difficult to determine which course (Borja 2007; Marceau 2008;
of action will be the most precise Toppeta 2010; Washburn,
Sindhu, Balaouras, Dines, Hayes, critical factors that
& Nelson 2010). [23, 24, 25, 26] These form pillar for the development of
problems trap the growth of city a smart city. This research paper
and dismantle explores various aspects and
it completely (Dawes, Cresswell & dimensions and
Pardo 2009; Rittel & Webber helps in bridging this gap by
1973; Weber & Khademian 2008) proposing a framework to give a
[27, 28, 29]. clear insight about the
This has triggered the usage of establishment of smart cities.
technology as a solution of all The framework proposed also
these issues and to address them helps in identifying recent trends
in a smarter way. and requirements for a city to
Hence the concept of Smart Cities evolve into a ―smart‖
is coming up. Smart Cities ensure city.
sustainable environment with the Ensuring sustainable development
help of Big and quality of life in complex
Data and Internet of Things. By social ecosystems of cities and
―smart‖, we mean that the city is urban areas are
more sustainable, livable and important concerns. Cities are
efficient. The smart increasingly aware of the concept
city market is estimated to reach of ―smart city‖ and are actively
an annual spending of about $16 developing
billion by the year 2020 (Pike strategies towards the goal of
Research 2011) [30]. becoming "smart" and manage city
Smart Cities, with a proper resources more efficiently while
roadmap, will serve people at large addressing
and will surely help in reducing development and inclusion
man power for the challenges.
long term. Though, the concept of To understand the concept of
smart city is still foggy in the smart city, we begin with
minds of people, few studies have understanding the conceptual
attempted to relatives of the model.
address the questions regarding Although limited in scope, they set
the development of smart cities. the theoretical framework for the
After performing an extensive holistic notion of the smart city
research in diverse currently
domains like public governance, understood by the urban planning
information technology, sphere. The genesis of the concept
egovernance, lie in Dutton‘s wired city, which
we have identified six promised to
use emerging telecommunications (Komninos 2008).
technology to provide Based on the exploration of a wide
unprecedented amounts of data and extensive array of literature
and information to from various disciplinary areas,
households and businesses the identified
through ―information highways‖ factors– Social, Management,
that, would create a Economy, Legal, Technology,
communications-centric society. Sustainability, (SMELTS) forms
Another precursor to the smart the basis of an
city is the digital city, a integrative framework. This
technologically-defined city that comprehensiveness is the
uses widespread broadband distinguishing factor of the smart
infrastructure to support e- city, which integrates a
Governance and ―a global number of physical, institutional,
environment for public and digital components to create a
transactions‖ (Mitchell, 2000). holistic definition of what smart
The notion of smart city is planning would
established from the combination look like.
of the knowledge society and In 2009, the number of people
digital city. It is defined as a living in urban areas (3.42 billion)
―multi-layer territorial system of surpassed the number living in
innovation‖ made up of digital rural
networks, individual intellectual areas (3.41 billion), and since then
capital, and the the world has become more urban
social capital of the city, which than rural. In 2014, there were 7.2
together constitute collective billion people living on the planet
intelligence (Komninos, 2008) —. (United Nations, 2014) [31]. It is
Economic estimated that by 2017, a majority
competitiveness and innovation of
achieved through the people were living in urban areas.
knowledgebased The global urban population was
economy marks a city as expected to grow approximately
intelligent, 1.84% per year between 2015 and
allowing it to generate a ―spatial 2020, 1.63% per year between
competitive advantage‖ through 2020
industrial districts, regions, and and 2025, and 1.44% per year
learning clusters between 2025 and 2030 (World
that produce sophisticated R&D Health Organization, 2014) [32].
and are supported by digital The increasing population trend
networks and artificial intelligence
shows us the importance of the city stakeholders an idea about
arranging how each smart city is making
city resources. Smart city projects progress. The results also help
are stakeholders make decisions; for
one of the efficient solutions. The example, investors may decide
use which
of smart computing technologies smart city project to invest in
makes the critical infrastructure based
components and services of a on a reliable ranking result. Many
city— companies, research institutes,
which include city administration, and
education, healthcare, public Non-Government Organizations
safety, (NGOs) are working on smart city
real estate, transportation, and ranking or evaluation (Albino,
utilities—more intelligent, Berardi,
interconnected, and efficient and Dangelico, 2015) [37]. They are
(Washburn et al., 2010) [33]. There typically displayed as a score or
is a ranking index.
range of conceptual variation This research will use fuzzy logic
generated by replacing smart with and machine learning techniques
other alternative adjectives such as to
digital city or sustainable city. predict whether a smart city will
Mills et be
al. (2022) [34] also give a definition classified as a leader or a follower.
of This research starts with the
smart city from the perspective of current
big smart cities‘ leader or a follower
data, artificial intelligence, and classification summary and
other analysis
characteristics. Oke et al. (2022) through fuzzy logic and machine
[35,36]
learning techniques. Based on the
found that all the smart city current smart cities classification
leaders result, some insighul rules and
and followers themselves can help information will be extracted for
each other to overcome some future smart cities prediction.
challenges. Only limited smart cities were in
Smart city ranking is a useful the prediction list, due to the
performance evaluation method. limitation of sampling framework,
There exist many smart city survey budgets, data accessibility,
ranking and
results. The ranking results give all
others. More cities should be have differentiated themselves
included in the prediction list in through the clarity, breadth, and
the inclusiveness of their smart city
future. Furthermore, different vision and planning. They are also
ranking leading the way in implementing
results use different significantprojects at both the
methodologies. pilot and increasingly full-scale
For example, an organization may levels.
use Smart City Follower: These are
survey methodology; another may cities that are beginning their
use secondary data. These smart city journeys. They may
differences have made initial statements of
lead to ―heteroscedastic‖ results. intent and begun
Based on the accessible smart limited pilot projects and soloed
city ranking results, a smart city operations, but they need to
can develop a more integrated view for
be either classified as a leader or a city development and/or stronger
follower. A fuzzy logic will be used leadership for their programs.
to Thus, the research question is
summarize the current smart city ―What machine learning algorithm
leaders and followers on the list. can accurately identify smart city
This leaders and followers based on
research paper applies several existing city indicators, and how
machine learning algorithms to can this knowledge be used to
identify smart city leaders and analyze smart city progress
followers by using some existing issues?‖.
city The research paper employs a
indicators. The highest test combination of fuzzy logic and
accuracy machine learning techniques to
algorithm will be used for identify and predict smart city
additional smart city leader and leaders and followers. The authors
follower predictions. Smart city first use fuzzy logic to label cities
progress issues will also be as either leaders or followers
investigated based on the based on evaluation meta results
prediction. from various organizations. They
In their assessment of smartest then apply machine learning
cities in the Gulf States, Woods et techniques to uncover the key
al. (2016) [38] define a smart city characteristics of each group.
leader and follower as follows: Using the Support Vector Machine
Smart City Leader: These cities (SVM) algorithm, the
authors use the training data‘s They observe that smart cities are
performance to predict which but one formulation of the more
cities are likely to become smart generic ‗future city‘ term, which is
city leaders or followers. The used to ―convey either
proposed prediction framework environmental, social, economic
successfully predicted 30 smart or governance aims, or a
city leaders and 20 followers. hybrid of some or all of these
Amidst the multitude of efforts elements‖ (p. 4). Additionally, the
surrounding the notion of the lack of cohesive understanding
smart cities, Hollands (2008) [39] may also be due to the various
formulates a critique on the usage different motivations that
of smart cities as a label. The call determine the choice of smart city
for clarification finds fertile soil in label. Cities gravitate towards
the research community, which concepts that are most appealing
assesses smart city research to be to them in that moment, which
fragmented, divergent, and lacking may be influenced by
unifying cohesion and intellectual factors such as geography and
exchange (Mora, Bolici, and zeitgeist (Eremia, Toma, and
Deakin, 2017) [40]. Hollands‘ Sanduleac, 2017) [42]. For example,
(2008) [39] main critique is that the aer the 1950s, the most popular
smart city label incorporates a term in urban development was
wide range of fields (from IT to ―sustainable city‖, while ―digital
business to communities). city‖ came up in the late 90s
However, it remains ambiguous in (Eremia et al., 2017) [42]. In 2009–
the ways in which these fields are 2010, ―smart city‖ became the
connected to the smart city notion dominant term with previously
and to each other. This is 132 documents published between
exemplified by the way that 2002 and 2009 to more than
―smart‖ can be replaced by a 900 in 2010–2012 (Mora et al.,
multitude of other adjectives, such 2017) [40]. The current discourse
as ―creative‖ or ―wise‖ cities, on future cities is distinctive for its
without increasing descriptive global, positive, strategic,
clarity. Although Hollands‘ integrated, and evidence-led
considers this overlap in meaning character (Moir et al., 2014) [41].
to be problematic, Moir, Moonen, This is also noted by Hollands,
and Clark (2014) [41] point out that who claims that the way that these
these slight differences may labels ―link together technological
indicate a desire to highlight one informational, transformations
of the specific Sustainability with economic, political and
aspects of the smart city concept. social-cultural change‖
(Hollands, 2008, p. 305) [10], commonsensical, and
which is generally positive in pragmatic. Still, he identifies
nature. With this positive several shortcomings that inhibit
connotation, cities are the growth of the smart city
generally eager to use these labels agenda. The first of
in an effort to appear more which is in line with Hollands‘
positive as well. Thus, a rhetorical argument that there is a lack of
inflation occurs in which the label shared understanding about the
loses its actual meaning and concept and initiatives. Kitchin
reference to technological and (2015) [45,46] then extends it by
infrastructural change in favor of claiming an overreliance on
marketing-fueled hype. This canonical and
conflation of labels also occurs simplified examples and an
with words that might initially absence
appear more neutral, such as of in-depth empirical case studies
―intelligent‖ or ―digital‖. These and
words similarly carry an optimistic comparative research in the
assumption regarding urban literature. In 2014, the European
development (i.e., a harmonious Parliament commissioned a report
high-tech future) and can have that maps the state of European
multiple possible meanings (see smart cities. To do this, they first
(Komninos, 2013) [43] for four outlined what a smart city seeks to
possible meanings of intelligent achieve (Manville, Europe,
cities). The purpose of Hollands‘ Millard, Institute, and Liebe, 2014,
paper was to break down the usage p. 17) [47]:
of the label and its assumptions, ―A Smart City is quintessentially
thus creating an opportunity for enabled by the use of technologies
other researchers (especially ICT) to improve
reflect on and seek clarification of competitiveness and ensure a
the notion of a smart city. For more sustainable future by
example, Allwinkle and symbiotic linkage of networks of
Cruickshank (2011) [44] people, businesses, technologies,
critically reflect on the concept of infrastructures, consumption,
―smartness‖ and other arguments energy and spaces‖. As such, their
set forth by Hollands. More working definition is (Manville et
recently, Kitchin (2015) [45,16] al., 2014, p.17) [47]: ―A Smart City is
contrasts Hollands‘ arguments by a city seeking to address public
arguing that the majority of the issues via ICT-based solutions on
smart city literature actually the basis of a multi- stakeholder,
appears to be nonideological, municipally based partnership.
These solutions are developed and and (8) the natural environment‖
refined through Smart (2012, p. 2291) [19]. Interestingly,
City initiatives, either as discrete the authors can on against using
projects or (more usually) as a these components to rank smart
network of overlapping activities‖. cites. Instead, they highlight these
2.2. Smart City Characteristics components as a supportive tool to
Since there is no commonly understand and advance smart
agreed-upon city strategies and initiatives. A
definition, substantial research similar approach was undertaken
effort is conducted on describing by Joshi, Saxena, Godbole
the characteristics of smart cities. and Shreya (2016) [50], who
The most prominent scheme propose a six-pillar framework
distinguishes six conceptually ―SMELTS‖: (1) social, (2)
distinct characteristics related to a management, (3) economy,
smart city: (4) legal, (5) technology, and (6)
(1) smart governance, (2) smart sustainability. In this framework,
people, (3) smart living, (4) smart technology, economy, and legal
mobility, (5) smart economy, and are said to have a greater impact
(6) smart environment (Giffinger, on and by the smart city
Fertner, Kramar, and Meijers, initiatives, which
2007) [48]. The European then affect the social,
Parliament follows this management, and sustainability
scheme in the sense that in factors in the outer level [51]. 2.3.
Sustainability 2023, 15, 9671 4 of Fuzzy Logic The core idea behind
17 order to qualify as a smart city fuzzy logic is that it aims
strategy or initiative it must to model the more imprecise
exhibit at least one of these six reasonings used by humans when
characteristics. they make rational decisions,
Other schemas approach the especially in an uncertain and
mayor from different perspectives. imprecise environment. This is
For example, Chourabi et al. possible due to the human ability
explored the literature from to
multiple fields to use imprecise, inexact, incomplete,
propose a framework containing or
eight core components of smart unreliable knowledge to infer an
city initiatives: ―(1) management approximate answer. Thus, fuzzy
and organization, (2) technology, logic
(3) governance, (4) policy, (5) seeks to extend logical reasoning
people and communities, (6) the in
economy, (7) built infrastructure, the sense that if logic is the
application of formal principles of particle physics to control
reasoning, then fuzzy logic is the engineering, robotics, and even
application of formal principles of natural languages.
approximate reasoning (Zadeh,
1998) [52]. Fuzzy logic is better At least five different descriptions
equipped to handle the concept of can be
a partial truth, because fuzzy logic encountered for ‗intelligent cities‘,
views everything, including truth as
itself, as a mayer of degree rather follows:
than a binary true or a. Intelligent cities are meant as
false. This does not mean that virtual
―fuzzy logic is fuzzy‖; rather, it is a reconstructions of cities or virtual
―precise logic of imprecision and cities
approximate reasoning‖. (Zadeh, (Droege 1997). The term has been
2008) [53]. Its principal facts are broadly
that it is logical, fuzzy-set used, covering a wide range of
theoretic, epistemic, and relational electronic
(Dzitac, Filip, and Manolescu, and digital applications relating to
2017) [54]. By providing a digital
mathematical means of spaces of communities and cities4.
representing vagueness, fuzzy b. World Foundation for ‗smart‘
logic models or sets are able to communities defines ‗intelligent‘
recognize, represent, manipulate, cities as
interpret, and utilize approximate ‗smart‘ cities which, based on the
information. This contrasts with adoption
more traditional Western and use of ICTs, are paving a
Aristotelian logic systems, which ‗smart‘
tend to be more binary in development. This implies a
approach. It initially drew mixed conscious
reactions as science, and effort to use ICTs to transform life
engineering at the time did not and work, within a certain region
consider the dullness of class (California Institute for Smart
boundaries [55]. Yet, Communities 2001).
the way that fuzzy logic seeks to c. ‗Intelligent‘ cities were also
formalize the human ability to defined as intelligent
reason and decide in situations of environments with embedded
imperfect importation is one of the ICTs, targeting the creation of
factors that has enabled fuzzy logic interactive spaces that bring
to be applied to many fields, from computation into the physical
artificial intelligence and quantum
world. (Steventon and Wright
2006; Bakis 2010).
d. Along the same lines,
‗intelligent‘ cities (communities,
clusters, regions) were defined as
multi-layer territorial systems of
innovation that bring together
knowledge intensive activities,
institutions for cooperation in
learning and innovation, and
digital spaces for communication
and interaction, in order to
maximize the problem-solving
capability of the city. The
distinctive characteristic, in this
respect, is highly innovative
performance, as
innovation and solving new
problems are the main features of
intelligence (Komninos
2002; 2006).
Methodology
This review aims to explore how
artificial intelligence is being used in the
modern smart city. We found no viable
review paper on AI adoption in smart
cities after thoroughly examining
relevant literature. This section
discusses the study's methodology by
proposing a conceptual framework for
AI uses in smart cities. Fig. 5 depicts the
conceptual framework of AI used in
smart cities. As shown in the
illustration, we selected common smart
city applications for the study. Fig. 5. Conceptual smart city-AI
framework.
We used keywords and keyword
combinations such as Artificial
Intelligence/AI, Machine Learning/ML,
Internet of Things/IoT, Smart
Healthcare, Smart City/Smart Cities,
Smart Education, Smart Infrastructure,
Smart Living, Smart Security, and
Smart Governance to search for
published papers in peer-reviewed
journals, international conferences, and
books in electronic bibliographical
sources such as Web of Science (WoS)
and Scopus. The initial search yielded
540 publications using the above
keywords. We studied all 249 related
main content and selected 133
publications (Scopus: 97%, WoS: 73%)
that are extremely relevant to our study
(publications between 2014 to 2021).
The publications included
computational and analytical
approaches, conceptual models, design
science, and case studies. Fig. 6 depicts
the proposed research methodology for
the source selection.
 3). Table 3 shows the number of
documents by year.

Year Total
2011 0
2012 1
2013 1
2012 4
2015 14
2016 28
2017 35
2018 63
Fig. 6. Proposed research methodology. 2019 92
The following are the inclusion 2020 145
criteria utilized to select the publications Total 383
for the study:
 1. Studies on the smart city Table. The number of documents by
notion have been undertaken. year.
They addressed AI's role in
several sectors, including This study conducted a systematic
education, healthcare, energy, review of smart city studies from the last
IT/IoT, mobility, security, ten years. The research question was set
localization, disaster and city as ‗What are smart cities‘ main issues
management, environment, and and research trends over the past ten
agriculture. years?‘ According to the research
 2. The research works appeared question, the analysis framework was set
in a peer-reviewed English as follows. In terms of research method,
journal, book, or conference year, research methodology 1, research
paper. methodology 2, research purpose, data
 3. The publication period is collection method, and use of research
between 2014 to 2021. results were included in the analysis
frame. First of all, in terms of the nature
In the case of academic journal searches, of the analysis data, the quantitative,
papers were collected for research from qualitative, and mixed research methods
journals using the keywords ‗smart city‘, used in Kim‘s research [25] [1] were
and by using ‗smart city‘ in Web of used. In research methodology 2, the
Science (Table 1). The selected papers case studies, literature studies, surveys,
include 85 in the SSCI-class and interviews used in Jeong and Yoo‘s
sustainability, 56 in the Sustainable research [26] [2] were used. The
Cities and Social Studies journals, 136 in research purpose of Kim‘s study [27] [3],
the SCI-class IEEE, and 106 in the to conduct an exploratory study,
Sensors journal (Table 2). A total of 387 descriptive research, and explanatory
papers were searched from 2011 to 2020 research, was also used. The data
during the research period, and three collection method used the data types,
papers without the original text and one such as primary and secondary, used in
withdrawn paper were excluded (Table Mok et al.‘s [28] [4] study. Finally, the
research results, basic research, applied
research, and evaluation research used analysed. Moreover, the classification
in Kim‘s study [29] [5] were used. In results are compared with the machine
addition, the analysis criteria included learning-based image classification
analysis units and classification of model.
research topics suitable for the
characteristics of smart city research. In 5.1. Data Construction
terms of research content, keywords Preparing data sets for object
related to research topics and topics,
detector was challenging due to the lack
analysis units, general classification
criteria for smart cities, and the of public datasets on drainage and gully
presence or absence of security were blockages. To overcome this, images are
included in the analysis frame. First of collected from publicly available image
all, regarding the subject and keyword, sources such as Google Image, YouTube
the topic modeling used by Park and Lee videos and other public sources. These
[30] [6] was used. As a result, this study images are retrieved using multiple
classified topics into keywords, such as ‗drainage block‘,
infrastructure/monitoring, ‗rubbish and drain blockage‘ and ‗drain-
citizens/sustainability, big grate and flooding‘. It was observed that
data/algorithms, smart grids, the all the collected images were not useful
internet of things/cloud, governance, considering our problem domain, owing
and transportation. Next, keywords in to some being noisy, unwanted and
the study were extracted and classified
blurred images. To remove those
for each subject. The analysis unit was
unwanted, noisy and blurred images
subdivided from micro-level technology
to macro-level countries and local from the dataset, manual supervision
governments. Nam and Pardo‘s [9] was applied for dataset preparation.
smart city classification criteria During the manual supervision, each
(technology and infrastructure, institute, image was analysed in terms of their
and human resources) were used for the size, quality and how closely they are
general classification of smart cities. In related to the problem domain. After
line with the recent trend of smart cities data pre-processing, the image dataset
in which informatization is progressing for each object type was prepared.
due to the active introduction of ICT
technology, this study analyzed smart 5.2. Image Augmentation
cities‘ research in terms of security Convolutional Neural Network
(cybersecurity and privacy).
(CNN) model with higher accuracy,
In this section, preparation of lower training and validation loss are
training data for the image coverage required. To build an effective CNN
detectors model, implementation details model, a larger training dataset is
using machine learning and the use of required [116 8.]. However, there has
semantic rules for image classification is been a limitation on accessing a large
presented in detail. Performance of the number of images for model training. To
hybrid classification model in terms of achieve this, image augmentation has
accuracies on object coverage detection proven to be a powerful technique to
and image classification are also enrich the image dataset. In other
words, image augmentation has found
to be an effective alternative to enhance 5.3. Image Annotation and
model performance [67 7,117 9.,118 Coverage Level
10.]. Supported by the reported
One of the major issues on object
improvement in model performance coverage detection is to find the
using image augmentation in the coverage proportion of each object
literature, we have chosen to use image within the image. The object count
augmentation to build an improved [120,121] method has been applied as
CNN model for object coverage detector an approach to find the density or
accuracy. proportion of object area within an
The main objective of applying image. However, the object count
image augmentation is to increase the approach for small objects appearing in
training image dataset. Different image a group has been a limiting factor for
augmentation techniques such as object count analysis [122]. In this
geometric transformation, random work, the object count method is not
erasing, colour space transformations viewed as a feasible option, as leaves
and feature space augmentation have and plastic and bottles are small in size
been applied for image augmentation. and often appears as a group.
―ImageDataGenerator‖ class from Keras Additionally, water and mud cannot be
library has been used as an alternative counted in discrete numbers. The object
for image augmentation to improve the count method also does not
image classification accuracy [69,119]. appropriately address the coverage area
Images are augmented by changing proportion. This can be illustrated with
augmentation parameters such as zoom, the example of analysing coverage of
height shift, width shift, shear and leaves as shown in Figure 7a,b. In these
brightness by altering the values as two figures, both the images have
shown in Table 1. Random values are approximately 53 % leaves coverage of
set within the range for different leaves. However, it can be observed that
parameters to increase the diversity in there is a considerable difference in the
image dataset. By applying image number of leaves in each image.
augmentation, the number of training Furthermore, it has also been visually
images was increased by approximately observed in those figures that there is no
10 times. uniform pattern on image coverages.
The size and shape of the objects vary
Table 1. Image Augmentation from image to image as well. Therefore,
parameters and value range. object counts within the image would
not be a feasible option to categorise the
image into different levels.
Figure 7. (a,b). Leave Coverage within images into different levels. To find the
the image, indicating the same coverage, coverage percentage of each object
but with a different count of objects. within the images, the area of each box
has been calculated for each object
To overcome these complexities,
based on box coordinates. The areas of
image annotation techniques have been
individual objects are summed, and
applied that mark the object of interest
hence, the percentage of coverage of
within the images. Using the annotation
each object within the image was
tool ―labellmg
calculated. Based on the coverage
(https://github.com/tzutalin/label
percentage, as listed in Table 2,
Img)‖, images were manually
annotated images have been categorised
annotated; an example of an annotated
into four coverage class levels. Example
image is shown in Figure 8. During the
of coverage level corresponding
annotation, boxes (boundaries) are
coverage percentage of three images,
created by covering the objects in the
shown in Figure 9a–c, is listed
image. There are one or more than one
in Table 3.
boxes for individual objects within the
image depending on the position and
orientation of the objects. Mostly, each
object covers some proportion of the
total area within the image. However,
there are overlapping boxes, mainly with
Figure 9. Sample Images of Different
the presence of water within the image.
Coverage Level (a): Two, (b): One, (c):
Two.
Table 2. Coverage level and coverage
percentage.

Table 3. Example of Object Coverage

Figure 8. Screenshot of ―Image Percentage and Corresponding Level.
Annotation‖.
The next challenge was to
categorise the annotated image into
different coverage levels. An analogy of
the object count method within the 5.4. Coverage Detector
frame, used for density calculation, has Implementation
been adopted to define the coverage
level [123]. The coverage area Object coverage detectors have been
percentage was applied to categorise the modelled, trained and implemented to
detect the coverage level of each object The main purpose of applying pooling
within the image. Coverage detector layers is to minimise the spatial
model for each object, leaves, plastic and resolution of the feature maps. Max
bottle and mud and water, has been pooling was implemented to
designed as a CNN model. progressively reduce the spatial size of
the feature maps, and hence, reduce the
5.4.1. Convolutional Neural
number of parameters and computation
Network
in the network.
CNN is a feed-forward neural
network made up of layers with Flatten Layer: a flatten layer
transforms the entire pooled feature
learnable parameters including weights
map matrix into a single column. The
and biases. A CNN is typically composed
transformed values are then fed to the
of multiple layers interconnected to each
next layer in the network for further
other. Different layers that have been
processing. Flatten layers are applied to
applied in CNN models are summarised
increase the training accuracy.
as below.
Dense Layer (a fully connected
Convolutional layers: Convolution
layer): several convolutional pooling
layer is the first layer in the CNN model
layers and flatten layers are applied in
and works as a feature extractor to
combination to extract features. The
extract features, such as edges or
dense layer that appears after these
textures from the input image. This
layers infers the feature representations
layer uses a convolution operation on
and perform the reasoning. In this work,
the input image and passes the extracted
output to the next layer in the model. the dense layer classifies the input image
into four levels. Softmax layer has been
The three-dimensional input in terms of
height, width and the number of used to calculate the final probabilities
channels are applied to the of each class.
convolutional layer to get the output as a Activation function: the activation
feature map. In this work, the number of function decides whether or not a node
convolution layers was adjusted based in the network should be activated. The
on the model training accuracy. After activation function introduces non-
several iterations and adjustment of linearity to the network, which analyses
model parameters and based on the best whether the information that a node is
training accuracy, two convolution receiving should be stored or ignored.
layers was set for leaves and water ReLU function was applied as the
coverage detector CNN models, whereas activation for the CNN models.
three convolutional layers were set for
plastic and bottles and mud coverage Four CNN models are designed by
detector CNN models. altering the number of layers. The CNN
model architecture, for the leaves
Pooling layers: a pooling layer is a coverage detector, is shown in Figure
down-sampling layer, which is 10. Other coverage detectors have
commonly applied after the convolution differed only in terms of the number of
layer and takes feature maps as its input.
convolution and pooling layers that were generated using data augmentation, as
used. discussed in Section 4.2.
Padding: padding is mainly applied
to add columns and rows with zeroes to
keep the spatial sizes the same after
convolution. Padding has been applied
such that the output has the same length
as the original input.
Figure 10. Convolutional Neural
Network (CNN) model (Leaves). 5.5. Semantic Representation

5.4.2. Model Regularisation and Semantic representation of the

Parameter Selection blockage instances was created after the
analysis of an individual expert‘s view
One of the most common problems on object identification, classification
in CNN training is overfitting. and reasoning. Structured knowledge is
Regularisation has been applied as one generated to represent the conceptual
of the approaches for improvement in model on blockages of drainage and
CNN training accuracy [124]. gully. The relationship between objects
Regularisation modifies the model‘s coverage level and classification
learning parameters such that the instances: ―Fully Blocked‖, ―Partial
performance of the model improves Blocked‖ and ―No Blockages‖ instances
during models training. In this work, have been presented with semantic
different parameters have been applied representation, as shown in Figure 11.
for the model regularisation: The concept map reflects the top-level
L2 (lambda): L2 is also known as concept of drainage and gully blockages.
weight decay. It has been applied so that This representation shows the major
the weights concentrate toward zero. components of an image in a certain
Lambda regularisation parameter value combination, which defines the
is set as 0.001. instances of blockages.

Dropout: dropout randomly

removes the output of some of the
hidden nodes during the training and
only updated the remaining nodes. We
have defined 0.25 to 0.5 as the
probability of dropout during the
training. After analysing different
ranges, a dropout probability of 0.4 was
set for the model, as it shows the best
training accuracy. Figure 11. Concept Map with the top-
level concept related to Blockage.
Image Augmentation: High number
of training image dataset has been ―Fully Blocked‖, ―Partially Blocked‖
and ―No Blockage‖ instances are
formulated based on objects and their classified with the ―No Blockage‖
coverage within the image. Using the category when water can freely flow into
experts‘ reasoning in image the drainage system and no obstruction-
classification into corresponding class causing objects are detected.
labels, the semantic rule-base has been Considering that the flooding condition,
created. The drainage and gully the experts have also highlighted that
blockages‘ instances have been the drains and gullies do not always
converted into the OWL (Web Ontology cope with the amount of water, but also
Language) ontology as shown in Figure the location and rain duration are
12. among the parameters that define the
flooding scenarios. In this hybrid image
classification model, the location and
rain duration parameter is ignored,
since with image analysis these
parameters cannot be detected.
Mutually exclusive semantic rules are
defined to classify the image instances
based on object coverage detection using
experts knowledge of image
classification. SWRL (Semantic Web
Rule Language) rules have been created
for all the possible instances depending
on the presence of object coverage and
Figure 12. OWL ontology of drain and their combinations; an example of one
gully blockage. of such rules are presented in Figure
13. We discuss the rules in more detail
5.6. Rule-Based Formulation in Section 6.2.
Experts have highlighted that an
image is classified as ―Fully Blocked‖
when there are many objects that cover
most of the image portion with the sign
of severe restriction of water flow
through the drainage system, i.e., water
cannot enter the drainage system due to Figure 13. Sample of SWRL rule
objects that caused a blockage. The research area surrounding the
Similarly, an image is classified into the concept of smart cities is very dynamic
―Partially Blocked‖ label when the water and in
is hindered from entering the drainage continuous evolution. This has resulted
system but is still able to enter to some in a vast knowledge being produced that
degree. In a drainage system with this allows
classification, there are fewer objects for an integrative literature review
that block it, or the combination of methodology. Hence, a careful collection
objects are not as severe. An image is and filtering of
relevant academic papers has been After an exhaustive evaluation of all the
performed for the scope. Two databases retrieved academic papers, additional
have been used: secondary
Google Scholar and the Catholic data (e.g., reports, websites of
University of the Sacred Heart‘s Online governmental agencies, newspaper
Public Access articles from
Catalogue (OPAC). The primary focus reputable publishers . . . ) were searched
was to collect peer-reviewed papers for on Google to further support the
published by emerging
high-quality publishers. The aim was to findings and to build up the case studies
retrieve studies that were pertinent to presented in Section 6. A deductive
previously approach has
formulated questions: been used to derive the final
 What studies concern the considerations.
development of AI in the three Finally, with regard to the other existing
proposed energy-related literature, this work aims to present an
areas (smart grid, EVs, and smart objective analysis of AI, avoiding
building) of a smart city? approaches that are either over-positive
 What research has been carried [14] 11 or biased by
out in relation to the negative anthropocentric interpretations of AI,
effects that AI can have when where the latter would be regarded as a
applied in those areas? convoluted
 How effective have nudging issue and potentially dangerous [15,16]
practices empirically been in 12,13.
inducing energy-saving behavior 2.1. The Concept of ―Smart City‖ and the
or in switching to renewable Contribution of This Research
energy? In the scientific literature, the
 Is AI a sustainable technology? assortment around the definition of a
What about data collection smart city (Table 1)
techniques? has contributed to ideals and
The references of the collected papers expectations that lack a corresponding
were used to find additional resources to concreteness in the real
expand world [17]. 14 Albino et al. [18] 15 have
the analysis. The inclusion criteria were supported the idea that the uniqueness
based on the relevancy of an academic of the objectives
paper in that may be endorsed may compromise
comparison with the issues planned for a universal assessment of a smart city.
investigation, with the objective of With regard
providing a to this matter, Burns et al. [19] 16 have
good balance between empirical discussed how the variety of smart city
research and qualitative studies. initiatives
Instead, exclusion was around the world has produced a term
considered for research with repetitive that is strictly related to the contexts in
results or for research that could not be which
contextualized it is deployed.
in the development of a smart city.
Source Definition
Harrison et al. ―Connecting the years, as it applies
[4] physical advances in the
infrastructure, the Internet of Things
IT infrastructure, (IoT)
the technology to
social enhance the
infrastructure, and quality and
the business efficiency of
infrastructure to services and
leverage resources‖
the collective Wang et al. [6] ―The idea . . . is to
intelligence of the use information
city‖. technology to drive
Almirall et al. ―A concept that the
[20] 17 encompasses most operation of the
of the areas where city, which
local includes
governments monitoring,
operate: forecasting, and
transportation, real-time
civic management. The
entrepreneurship, combination of IoT
democratic and AI can
transparency, replace the
clean energy, and traditional means
services provision‖ of managers in the
Mohanty et al. ―A place where past.‖
[7] traditional Lazaroiu and ―The large and
networks and Roscia [22] 19 small districts are
services are made proposing a new
more city model,
flexible, efficient, called the smart
and sustainable city, which
with the use of represents a
information, community of
digital and average
telecommunication technology size,
technologies, to interconnected
improve its and sustainable,
operations for the comfortable,
benefit of its attractive and
inhabitants‖. secure.‖
Park et al. [21] ―A concept has
18 gained substantial Table 1. Some definitions of the term
attention over the ―smart city‖ in the scientific literature.
last few
The study of Ahvenniemi et al. [23] 20 is work of Ranchordás [30] 27 with regard
among the first to highlight the to the academic literature on smart
importance of cities in terms of
the sustainable development of smart legal and ethical issues. Hence, a review
cities and how the employment of new of the effectiveness of hypernudge
technologies techniques
should not be an end in itself. In aimed at conscious energy consumption
particular, from this perspective, will be provided to expand this
Kramers et al. [24] 21 discussion.
have assessed how different ICT
solutions could be used to reduce energy 1] Type Of Research Conducted: -
consumption in Research conducted in the
smart cities. More recently, Hoang et al. implementation of AI in smart cities
[25] 22 have reviewed the introduction spans various domains and can be
of renewable categorized into several key areas:
resources into smart cities, along with Urban Planning and
the potential challenges. As electric Infrastructure Optimization:
vehicles are Research in this area focuses on using
becoming an important topic in the AI to optimize the design and
discussion on the demand-side of a management of urban infrastructure.
smart grid, the This includes developing algorithms to
barriers to their adoption have been improve traffic flow, reduce energy
comprehensively evaluated by Adhikari consumption, and enhance water
et al. [26] 23 in connection with the resource management. Machine
energy requirement. Similarly, the learning models can be employed to
argument has been assessed by predict future infrastructure needs and
Sanguesa et al. [27] 24, but in the plan accordingly.
specific framework of smart cities. While Smart Mobility: AI is used to enhance
many AI applications transportation systems within smart
have been developed for the smart grid, cities. Research may involve developing
and Omitaomu and Niu [28] 25 have autonomous vehicles, intelligent traffic
provided management systems, and ride-sharing
a complete survey on the matter, there is optimization algorithms. AI can also
yet to be an analysis of whether the facilitate the integration of various
introduction transportation modes, such as buses,
of AI can be an actual solution to trains, and bicycles.
environmental concerns. In fact, Energy Management: AI helps in
Yigitcanlar et al. [9] have optimizing energy consumption in smart
provided the general shortfalls of AI cities by predicting demand patterns
when deployed in smart cities. It is and adjusting energy distribution
important to point accordingly. Research in this area
out that data collection can not only includes smart grid management,
support AI but also nudging techniques. demand response systems, and energy-
The idea of efficient building management systems.
hypernudging has been presented by Environmental Sustainability: AI
Yeung [29] 26, and it has been further can contribute to environmental
enriched by the sustainability by monitoring air quality,
predicting pollution levels, and AI in smart cities, including its effects on
managing waste more efficiently. employment, inequality, and quality of life.
Research in this area focuses on creating This research helps policymakers make
AI-powered sensors and monitoring informed decisions.
systems. Interoperability and Standardization:
Ensuring that different AI systems in a
Public Safety and Security: AI-
smart city can communicate and work
driven solutions are used to enhance together effectively is a key challenge.
public safety and security in smart cities. Research involves developing standards and
Research may involve developing protocols to promote interoperability.
predictive policing models, surveillance The implementation of AI in smart cities is
systems with facial recognition, and an interdisciplinary field, involving
emergency response optimization. computer science, urban planning,
Healthcare and Wellness: AI can engineering, social sciences, and more.
play a significant role in improving Researchers collaborate across these
healthcare services within smart cities. disciplines to address the complex
Research may include developing challenges and opportunities presented by
smart city technologies.
healthcare chatbots, remote patient
2. How Data is collected and analysed:
monitoring, and disease outbreak -
prediction systems. Collecting data to analyze the use of AI in
Data Analytics and Insights: Smart smart cities can be done by a normal
cities generate vast amounts of data, and person, even without specialized training or
AI is used to analyze and extract resources. Here are some steps you can
meaningful insights from this data. follow to collect data for such an analysis:
Research in this area involves developing Online Research: Start by conducting
data analytics tools, predictive models, and online research. Look for publicly available
visualization techniques to support reports, articles, and whitepapers related to
decision-making. AI implementations in smart cities.
Citizen Engagement and Services: AI- Government websites, academic journals,
driven chatbots and virtual assistants can and technology news outlets can be valuable
enhance citizen engagement and provide sources of information.
efficient public services. Research may focus Case Studies: Seek out case studies of
on creating user-friendly interfaces and specific smart city projects that have
natural language processing capabilities for implemented AI technologies. These case
interacting with citizens. studies often provide detailed information
Ethical and Regulatory on the goals, challenges, and outcomes of AI
Considerations: Given the sensitive initiatives in smart cities.
nature of AI applications in smart cities, Surveys and Interviews: If possible,
research includes studying ethical and legal conduct surveys or interviews with
aspects. This involves addressing privacy residents, city officials, or experts in the
concerns, data security, and transparency in field. This primary data can provide insights
AI systems. into the real-world impact and perception of
Human-Centric AI: Research in this area AI in smart cities.
aims to make AI more user-centric and Public Data Sources: Many smart cities
accessible to all citizens. This includes publish open data sets related to their
developing AI systems that are inclusive, initiatives. You can explore these datasets to
easy to use, and considerate of diverse user gather information on topics such as traffic
needs. patterns, energy consumption, air quality,
Socioeconomic Impact: Researchers and public services. Websites like data.gov
also investigate the socioeconomic impact of can be a valuable resource.
Social Media and Forums: Monitor gather valuable data to analyze the use of AI
social media platforms, forums, and in smart cities and gain a better
community discussions related to smart understanding of the subject.
cities and AI. These platforms often contain
valuable insights, user experiences, and
opinions.
Attend Local Events and Workshops:
Participate in local smart city events,
conferences, and workshops if possible.
These events can provide opportunities to
network with experts and gather firsthand
information.
Government Reports and
Publications: Government agencies often
publish reports on smart city initiatives,
including details on AI projects. Check the
websites of municipal, regional, or national
government bodies for relevant
publications.
Academic Research: Explore academic
research papers and theses related to AI in
smart cities. Universities often conduct in-
depth studies on various aspects of smart
city technology.
Industry Reports: Look for industry
reports and market research related to AI in
smart cities. These reports often provide
data on market trends, investment, and
technology adoption.
Collaborate with Local Organizations:
Collaborate with local community
organizations or nonprofits that focus on
smart city initiatives. They may have access
to data or insights that can be shared.
Online Courses and Webinars:
Consider taking online courses or attending
webinars related to smart cities and AI.
These resources can provide you with a
deeper understanding of the topic and may
offer access to additional data sources.
Join Online Communities: Join online
communities, forums, or LinkedIn groups
related to smart cities and AI. Engaging
with experts and enthusiasts in these
communities can help you stay updated on
the latest developments and data sources.
Remember to verify the credibility and
reliability of the sources you use. Cross-
reference information from multiple sources
to ensure accuracy. By combining
information from various sources, you can
 Implementation Of Work
Topic: - Application of Artificial Intelligence in Smart Cities

General Application of AI in smart cities: -

AI plays a vital role in the development and operation of smart cities, enhancing various aspects
of urban life. Here are some key applications of AI in smart cities:
Traffic Management: AI optimizes traffic flow by analyzing data from various sources, such as
cameras, sensors, and GPS systems. It helps in predicting and managing traffic congestion,
optimizing traffic signals, and suggesting alternate routes to reduce commute times.
Public Safety and Security: AI-powered systems monitor video feeds from surveillance cameras
to detect and respond to unusual activities in real-time. It can identify potential security threats,
recognize patterns of criminal behavior, and help law enforcement respond more effectively.
Waste Management: AI enables more efficient waste collection by optimizing routes and
schedules based on real-time data. Smart waste management systems use sensors to determine
the fill-level of bins, improving collection efficiency and reducing operational costs.
Energy Management: AI helps in the efficient use of energy resources by predicting demand,
optimizing energy distribution, and identifying areas for energy conservation. Smart grids,
enabled by AI, help balance energy supply and demand.
Healthcare Services: AI can improve healthcare access and services by analyzing healthcare
data, predicting disease outbreaks, and optimizing resource allocation. Additionally, it can help
in monitoring public health and identifying potential health risks.
Environmental Monitoring: AI technologies facilitate monitoring air and water quality,
identifying pollution sources, and predicting environmental changes. This data aids in
implementing policies to mitigate environmental risks.
Smart Governance: AI assists in data-driven decision-making for city planning, policy
development, and resource allocation. It can analyze vast amounts of data to identify trends,
preferences, and patterns, aiding city administrators in making informed decisions.
Public Transportation: AI optimizes public transportation systems by predicting demand,
managing schedules, and improving routes to increase efficiency and convenience for
commuters.
Smart Infrastructure Maintenance: AI helps in predictive maintenance of city infrastructure,
such as bridges, roads, and public facilities. It predicts potential failures, reducing downtime and
the cost of emergency repairs.
Citizen Services and Engagement: AI-powered chatbots and virtual assistants improve citizen
services by providing 24/7 support, answering queries, and offering information on various city
services and events.
The integration of AI in smart cities leads to increased efficiency, cost savings, better resource
management, and improved quality of life for residents. However, ethical considerations, data
privacy, and security challenges must be addressed for the successful implementation of AI in
smart city initiatives.

Model for Application of AI in field of healthcare: -

Note: - Since AI has vast implementation in smart cities across different fields, here we are
taking only one field into consideration that is healthcare.
Model: - Creating a predictive health analytics system for hospitalized patients involves various
steps and considerations. Here's a basic Python example using a machine learning approach to
predict the risk of certain health outcomes based on patient data. This example will use the
Scikit-learn library for simplicity.
Note: This is a basic example for illustrative purposes and might not cover all aspects of a
complete predictive health analytics system.
Formula Used by the Predictive Healthcare Analytics for analyzing the stability in condition of
the patient: -
Healthcare predictive analytics models utilize various statistical and machine learning algorithms to
analyze data and make predictions or recommendations. The specific formula or algorithm used depends
on the nature of the problem being addressed. Here are some common techniques and formulas used in
predictive healthcare analytics models:

1. Linear Regression: A basic formula used to predict a continuous outcome variable based on one
or more predictor variables. The formula for simple linear regression is:
𝑦̂ = 𝑏0 + 𝑏1 𝑥 where 𝑦̂ is the predicted value, 𝑏0 is the intercept, 𝑏1 is the slope, and 𝑥 is the
input variable.

2. Logistic Regression: Used for binary classification problems (e.g., predicting whether a patient
1
has a certain condition or not). The formula for logistic regression is: 𝑃(𝑦 = 1) = 1+ 𝑒 (𝑏0+𝑏1𝑥)
where 𝑃(𝑦 = 1) is the probability of the positive class, and 𝑏0 , 𝑏1 are coefficients.

3. Random Forest: An ensemble learning method that operates by constructing multiple decision
trees and outputs the class that is the mode of the classes (classification) or mean prediction
(regression) of the individual trees.
 4. Support Vector Machines (SVM): A supervised learning model used for classification or
regression analysis. In classification problems, the algorithm finds the hyperplane that best
separates data into different classes.

5. Neural Networks: Complex models that can learn non-linear relationships. The formula includes
various activation functions, layers, weights, and biases.

6. K-Nearest Neighbors (KNN): A simple algorithm that stores all available cases and classifies new
cases based on a similarity measure.

7. Decision Trees: Models that recursively split data based on features to make decisions. They are
based on a series of if-else questions.

8. Time Series Forecasting Models: Such as ARIMA (Autoregressive Integrated Moving Average) or
Exponential Smoothing models used for forecasting future values based on historical data.

These models can be further enhanced by feature engineering, hyperparameter tuning, and validation
techniques to improve accuracy and generalizability. It's important to note that while these formulas and
techniques are widely used, the choice of the most suitable model depends on the specific healthcare
problem, the nature of the data, and the available resources.

Additionally, ethical considerations, data privacy, and regulatory compliance are crucial in healthcare
analytics, affecting model development and deployment.

Source Code in Python for Implementation of above Model: -

# Necessary libraries
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import accuracy_score

# Sample patient data (you would need actual data for a real system)
# This data includes patient vitals, demographics, and health records
data = {
'Age': [45, 50, 60, 30, 65],
'Heart Rate': [75, 80, 90, 70, 85],
'Blood Pressure': [120, 130, 140, 110, 150],
'Cholesterol': [200, 220, 240, 180, 260],
'Diabetes': [0, 1, 1, 0, 1],
'Smoker': [0, 1, 0, 0, 1],
'Outcome': ['Stable', 'Unstable', 'Unstable', 'Stable', 'Unstable']
}

# Creating a DataFrame from the sample data

patients = pd.DataFrame(data)
# Defining features and target variable
features = ['Age', 'Heart Rate', 'Blood Pressure', 'Cholesterol',
'Diabetes', 'Smoker']
target = 'Outcome'

# Splitting the data into features and target variable

X = patients[features]
y = patients[target]

# Splitting the data into training and testing sets

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2,
random_state=42)

# Creating and training the model (Random Forest Classifier)

model = RandomForestClassifier(n_estimators=100, random_state=42)
model.fit(X_train, y_train)

# Making predictions
predictions = model.predict(X_test)

# Evaluating the model

accuracy = accuracy_score(y_test, predictions)
print(f"Model Accuracy: {accuracy}")

# Now, with a trained model, you can use new patient data to predict
outcomes
new_patient_data = {
'Age': [55],
'Heart Rate': [70],
'Blood Pressure': [125],
'Cholesterol': [210],
'Diabetes': [1],
'Smoker': [0]
}

new_patient = pd.DataFrame(new_patient_data)

# Making predictions for new patients

new_predictions = model.predict(new_patient)
print(f"Predicted Outcome for New Patient: {new_predictions}")
Result of the above Source Code: -

Conclusion: -

1. Model accuracy is 0 because our data is static and it has no value to compare with.
2. If all the vitals of the patient are within the borderline for that respective age then the model
gives ‘STABLE’ as output else ‘UNSTABLE’ .
Recommenda ons and Conclusion typically located in historical, centralized IT
This research paper indicates that smart architectures, where a main server, which is
cities need to embrace artificial often scalable only at the risk of very high
intelligence technology to improve their cost, elaborates the entirety of the incoming
performance. Apart from embracing data. Following the diffusion of the internet
on a global scale, the usual textual
artificial intelligence, smart cities should
representation of information has passed
adopt strategies to accommodate the
through numerous transformations,
technology within the numerous
becoming visual, vocal, and finally
functions to develop a municipality. Such iconographic, making its interpretation more
measures will ensure that the concept of complex. The exposure of the individual to
smart cities is recognized and adopted in the web and, more specifically, to an
most parts of the world. entirely wired smart city where whatever
Conclusively, the development of information can be measured is potentially
intelligent cities is promising to improve collected and where cameras monitor the
the standard way of life in urban centers flow of traffic or even the trajectories and
within the public sector and the private movements of pedestrians to prevent safety
sector. issues, presents many concerns in terms of
However, developing smart cities needs privacy and freedom. A potential side-effect
careful consideration of the technology to of over-information may result in algorithms
enhance their sustainability. Artificial built to nudge smart citizens—to preserve
intelligence plays a vital role in smart water and energy or to take public transport
cities by ensuring improved security, smart to reduce pollution—that are barely
policing, proper waste management, effective, due to a skepticism about the way
efficient energy use, and smart parking. information is circulated and handled by
In a smart city, governance is a local authorities. Cyberattacks and the
fundamental asset as policies improve the undisciplined circulation of data about
quality of life of the citizens, enhance health are other sources of concern.
leadership, protect the environment, and Firstly, more rational IT architectures
support local economies. However, one of have been adopted. They are based on
the challenges in planning effective distributed networks, where small nodes
practices in a smart city is uncertainty, an with limited processing capacity are
issue mitigated by collecting information interconnected. Furthermore, specific units
from heterogeneous sources and developing act as workload balancers, optimizing
proper algorithms dedicated to their energy consumption by distributing the data
analysis. The outcome of this process stream over the nodes in a weighted manner.
represents a valuable resource to support Secondly, with regard to the privacy
decisions on a rational basis. concerns, the European General Data
However, the cost of this operation Protection Regulation (GDPR) poses strict
should not be neglected, as the hardware and directives on data retention, data
the data storage facilities adopted by confidentiality, and data consent. Similar
software can have a strong impact from the regulations are discussed by the California
energy point of view and, ultimately, on the Consumer Privacy Act (CCPA), Brazil’s Lei
environment. With regard to information Geral de Proteção de Dados (LGPD) and
processing, the bottleneck—both in terms of South Africa’s Protection Of Personal
performance and energy consumption—is Information (POPI). In a similar manner, the
EU has recently adopted the nudging techniques. From this study, it has
recommendation on a European Electronic emerged that not all the nudging techniques
Health Record exchange format, though the are effective when the target is to reduce
threat of cyberattacks remains potentially energy consumption. However, these
harmful. specific nudges are already endowed with a
More concerns revolve around AI and, significant degree of personalization, with
specifically, the cost of deploying it targeting being inevitable as single
pervasively as this might result in further households or neighborhoods are usually the
controversies, i.e., (i) job losses, due to an addressees of such methods. This allows for
uncontrolled replacement of working the introduction of AI-backed nudges—also
activities considered repetitive and barely referred to as hypernudging. Yet, their
motivating; (ii) populism, deriving from success is closely linked with the personal
undetected bias in datasets used to train propensity of each individual. For instance,
algorithms that are supposed to guide the a free-riding tendency may compromise
individual in taking choices; (iii) ethical their impact, resulting in highly context-
concerns, for example when the individual dependent results.
has no role in decisions typically drawn by Continuing the discussion surrounding
an algorithm; and, finally, (iv) lawful issues, AI and the energy-related components, these
related to the lack of a formal definition of components themselves can be enhanced
AI as a legal identity. All these aspects can with the technology. The evolution towards
be mitigated by the definition of proper a smart grid has been strictly associated with
polices aiming at (i) supporting and the introduction of renewable resources and
encouraging AI-inclusive education on-site production that allows consumers to
programs; (ii) implementing algorithms have an active role. AI applications can be
alleviating and reducing bias in datasets, (iii) of crucial support but should not result in an
enforcing actions based on cooperation additional source of high energy
between humans and AI software; and (iv) consumption. Otherwise, a vicious circle
pursuing a robust definition of legal may be inadvertently created, with AI
personhood defining the rights and duties of instead becoming an additional
AI software. environmental issue. The electrification of
The results provided by the research vehicles is linked with a functioning smart
indicate that both data and AI can be grid. The introduction of electric vehicles
legitimately added to the pillars of smart should be coupled with the one of renewable
cities. energy, or it can result in indirect emissions
Recently, there has been a shift in the due to carbon-intensive resources. AI can
perception of the concept of a smart city, act as an aid in localizing areas where
and scholars are recognizing the importance charging points can be positioned,
of environmental awareness in its scheduling charging events and participating
development. From this perspective, the in the vehicle-to-grid system. However,
energy-related components (smart grid, negative effects should be considered in all
electric vehicles, and smart buildings) play a these applications, including the use of
fundamental role in the reduction in the carbon-intensive energy, the risk of
ever-increasing greenhouse emissions overloading, and the deterioration of
produced by cities around the world. With batteries made with scarce materials such as
regard to the proposed two pillars, lithium. Smart buildings are another
information is crucial to design adequate important element of the smart grid, as they
allow for the efficient demand-side drastically changing city life. Smart cities
management of energy. For the optimal are enhanced via computer vision and
management of the on-site production of artificial intelligence technologies, resulting
renewable energy, AI applications have in increased productivity, efficiency, safety,
shown to possess promising results. and economic benefi
Forecasting the energy demand and supply
can reduce the imported electricity from the Funding: This research received no external
grid, both when an energy storage system is funding.
present and when it is not. Nevertheless, the Institutional Review Board Statement:
installation of solar panels, wind turbines, Not applicable.
and storage technologies can be a Informed Consent Statement: Not
considerable investment that cannot be applicable.
widely embarked upon. Renewable Data Availability Statement: Not
resources are not the only way to solve the applicable.
problems of energy-related emissions in Acknowledgments: No external funding is
cities. AI-backed energy management aimed required for this study.
at reducing consumption is a viable solution. Conflicts of Interest: The authors share no
However, what must be considered is that conflict of inter.
people can easily interfere with it.
The results coming from the real-life
cases show that AI applications for energy-
related components are still in a
developmental phase. Hence, there is no
strong evidence that points against the
adoption of AI in this component of a smart
city. This should be taken as an opportunity
to avoid the drawbacks that are likely to
emerge with wider adoption.

Artificial intelligence (AI) bids serious

attention among the researchers and
academicians to show how the evolutions of
smart city are taking place at different
scales. Smart Cities and Artificial
Intelligence give a multidisciplinary, joined
procedure, using speculative and applied bits
of information, for the evaluation of savvy
city situations and citizens. The objective of
this published special issue is to focus on all
aspects and future research directions on the
specific area of AI-enabled applications for
automatic and intelligent systems in smart
cities and societies
Overall, Smart City technology's potential
applications hold great promise for
 biotechnology, artificial intelligence,
References
robotics. Control Engineering, 60 (n, a-n,
[1] Bernabéetal.,2015; Bibri,2018b; a).
Chengetal.,2018; Li et al., 2016; c. Kaplan, J. (2017). Viewpoint: Artificial
Neirotti et al., 2014a; Rolf, Pauleit, & intelligence: Think again. 60, Association
Wiggering, 2018 for Computing Machinery.
[1] a. Bernabé, A., Bernard, J., Musy, M.,
Andrieu, H., Bocher, E., Calmet,I., ... Rosantab
Communications of the ACM 36.
(2015). Radiative and heat storage properties of d. Wilson, H., Daugherty, P., & Morini-
the urban fabric derived from analysis of Bianzino, N.(2017). The jobs that artificial
surface intelligence will create. MITS loan
forms. Urban Climate, 12, 205–218. Management Review, 58, 14–16.
b. Bibri ,S.E.(2018a). A foundational framework [7] Bari, 2017; Fan, Han, & Liu, 2014; Tole,
for smart sustainable city development: 2013)
Theoretical, disciplinary, and discursive
[7] a. Bari, A. (2017). Working with big
dimensions and their synergies. Sustainable
Cities and Society, 38, 230–253. data: Scaling data discovery. West mouth,
c. Neiro, P. , DeMarco, A. , Cagliano, A.C. , Canada: Abdallah Bari.
Mangano, G., & Scorrano, F. (2014a). Trends in b. Fan, J., Han, F., & Liu, H. (2014).
smart city initiatives: Some stylized facts. Cities, Challenges of big data analysis. National
38, 25–36. Science Review, 1, 293–314.
d. Rolf, W., Pauleit, S., & Wiggering, H. (2018). A c. Tole, A. A. (2013). Big data challenges.
stakeholder approach, door opener for Database Systems Journal, 4, 31–40.
farmland [8] Koenig, Miao, Knecht, Bus, & Mei-
and multifunctionality in urban green
Chih, 2017
infrastructure (In Press)
[2] Levy, A. (1999). Urban morphology and the [8] a. Koenig, R., Miao, Y., Knecht, K., Bus,
problem of the modern urban fabric: Some P.,
questions for research. Urban Morphology, 3, & Mei-Chih, C. (2017). Interactive urban
79–85. synthesis: Computational methods for fast
[4] Alvalez, R. (2017) . The relevance of prototyping of urban design proposals.
informational infrastructures in future CAAD Futures, 724, 23–41.
cities. [9] O’Dwyer, E., Pan, I., Acha, S., & Shah, N.
Field actions science reports. (2019). Smart energy systems for
[5] Allam, Z. , & Jones , D. (2018a). Towards sustainable smart cities: Current
a circular economy: A case study of waste developments, trends and future directions.
conversion in to housing units in Cotonou, Applied energy, 237, 581-597.
Benin. Urban Science, 2, 118. [10] Liu, Y., Yang, C., Jiang, L., Xie, S., &
[6] Davenport & Ronanki, 2018; Hoske, Zhang, Y. (2019). Intelligent edge computing
2013; Kaplan, 2017; Wilson, Daugherty, & for IoT-based energy management in smart
Morini-Bianzino, 2017). cities. IEEE network, 33(2), 111-117.
[6] a. Davenport, T., & Ronanki, R. (2018). [11] Ullah, Z., Al-Turjman, F., Mostarda, L.,
Artificial intelligence for the real world. &
Harvard business review. Gagliardi, R. (2020). Applications of artificial
intelligence and machine learning in smart
b. Hoske, M.(2013). Hot technology jobs:
cities. Computer Communications, 154,
Computers, telecommunications,
313-323 Economic and Social Affairs, Population
[12] Petrolo, R., Loscri, V., & Mion, N. Division: the 2011
(2017). Towards a smart city based on cloud Revision: Highlights. 2012
of things, a survey on the smart city vision [21] Cisco Report, Smart Cities and Internet
and paradigms. Transactions on emerging of Everything-The Foundation for Delivering
telecommunications technologies, 28(1), Next-Generation Citizen Services,
e2931. sponsored by Cisco. 2013; Available from
[13] Aguilera, U., Peña, O., Belmonte, O., & hps://www.cisco.com/web/strategy/docs/
López-de-Ipiña, D. (2017). Citizen-centric scc/ioe_citizen_svcs_white_paper_idc_201
data services for smarter cities. Future 3.pdf
Generation Computer Systems, 76, 234- [22] Johnson, B. Cities, systems of
247. innovation and economic development.
[14] Al-Turjman, F. (2017). Innovation: Management, Policy & Practice,
Informationcentric 2008; 10(2-3), 146-155
sensor networks for cognitive IoT: an [23, 24, 25, 26]
overview, Ann Telecommun, 72(1), 3-18. [23] Borja, J. Counterpoint: Intelligent cities
[15] Allam, Z., & Dhunny, Z. A. (2019). On and innovative cities. Universitat Oberta de
big data, artificial intelligence and smart Catalunya (UOC) Papers: E-Journal on the
cities. Cities, 89, 80-91. Knowledge
[16] Li, H., Wei, T., Ren, A., Zhu, Q., & Wang, Society, 5. 2007; Available from:
Y. (2017, November). Deep reinforcement hp://www.uoc.edu/uocpapers/5/dt/eng/
learning: Framework, applications, and mitchell.pdf.
embedded implementations. In 2017 [24] Marceau, J. Introduction: Innovation in
IEEE/ACM International Conference on the city and innovative cities. Innovation:
Computer-Aided Design (ICCAD) (pp. 847- Management, Policy & Practice, 2008; 10(2-
854). IEEE. 3), 136-145
[17] Ramchurn, S. D., Vytelingum, P., [25] Toppeta, D. The Smart City Vision: How
Rogers, Innovation and ICT Can Build Smart,
A., & Jennings, N. R. (2012). Pung “Livable”, Sustainable Cities. The Innovation
the'smarts' into the smart grid: a grand Knowledge
challenge for artificial intelligence. Foundation. 2010; Available from
Communications of the ACM, 55(4), 86-97. hp://www.thinkinnovation.org/file/resear
[18] Allam, Z., & Newman, P. (2018). ch/23/en/Toppeta_Report_005_2010.pdf
Redefining the smart city: Culture, [26] Washburn, D., Sindhu, U., Balaouras, S.,
metabolism and governance. Smart Cities, Dines, R. A., Hayes, N. M., & Nelson, L. E.
1(1), 4-25. Helping CIOs Understand "Smart City"
[19] Srinivas, T. A. S., & Manivannan, S. M. Initiatives:
(2020). Preventing collaborative black hole Defining the Smart City, Its Drivers, and the
aack in IoT construction using a CBHA– Role of the CIO. Cambridge, MA: Forrester
AODV routing protocol. International Research, Inc. 2010; Available from
Journal of Grid and High Performance hp://public.dhe.ibm.com/partnerworld/p
Computing (IJGHPC), 12(2), 25-46. ub/smb/smarterplanet/forr_help_cios_und
[20] United Nations. World Urbanization _smart_city_initiatives.pdf.
Prospects. United Nations, Department of [27] Dawes, S. S., Cresswell, A. M., & Pardo,
T. A. From "need to know" to "need to share” Evidence-based public policy decision-making in
Tangled problems, information boundaries, smart cities: Does extant theory support
and the building of public sector knowledge achievement of city sustainability objectives?
networks. Public Administration Review, Sustainability 2022, 14, 3.
[35] AlAwadhi, S.; Scholl, H.J. Aspirations and
2009; 69(3), 392-402.
realizations: The smart city of Seale. In
[28] Riel, H. W. J., & Webber, M. Proceedings of the 2013 46th Hawaii
Dilemmas in a general theory of planning. International Conference on System Sciences,
Policy Sciences, 4(June), 1973; 155-169. Wailea, HI, USA, 7–10 January 2013; pp. 1695–
[29]. Weber, E. P., & Khademian, A. M. 1703.
Wicked problems, knowledge challenges, [36] Oke, A.E.; Stephen, S.S.; Aigbavboa, C.O.;
and collaborative capacity builders in Ogunsemi, D.R.; Aje, I.O. Smart City Team
network sengs. Public Partnership. In Smart Cities: A Panacea for
Administration Review, 2008; 68(2), 334 Sustainable Development; Emerald Publishing
349 Limited: Bingley, UK, 2022.
[30] Pike Research on Smart Cities [37] Albino, V.; Berardi, U.; Dangelico, R.M.
Smart cities: Definitions, dimensions,
[dedicates entire section to World sensing].
performance, and initiatives. J. Urban Technol.
[Online]. 2011; Available:
2015, 22, 3–21.
hp://www.pikeresearch.com/research/sm [38] Woods, E.; Omara, H.; Ravens, S.; Citron, R.
art-cities Gulf States Smart Cities Index: Assessment of
[31] United Nations. Department of Economic Strategy and Execution for 10 Cities (White
and Social Affairs, Population Division. World Paper); Navigant Research: Boulder, CO, USA,
Urbanization Prospects: The 2014 Revision, 2016.
Highlights (ST/ESA/SER.A/352). 2014. Available [39] Hollands, R.G. Will the real smart city
online: please stand up? City 2008, 12, 303–320.
hps://www.google.com.hk/url?sa=t&rct=j&q= [40] Mora, L.; Bolici, R.; Deakin, M. The first two
&esrc=s&source=web&cd=&cad=rja&uact=8&v decades of smart-city research: A bibliometric
ed=2ahUKEwjlxcPioML_AhWNU94KHbezCywQF analysis. J. Urban Technol. 2017, 24, 3–27.
noECBYQAQ&url=hps%3A%2F%2Fesa.un.org% [41] Moir, E.; Moonen, T.; Clark, G. “The Future
2Funpd%2Fwup%2Fpublications%2Ffiles%2Fwu of Cities: What Is the Global Agenda?” The
p2014-report.pdf&usg=AOvVaw1K8kVV-23- Business of Cities; UK Government: London, UK,
am3YUcFIpk4W 2014. Available online:
(accessed on 3 September 2023). hps://assets.publishing.service.gov.uk/govern
[32] WHO Expert Commiee on the Selection, ment/uploads/system/uploads/aachment_dat
Use of Essential Medicines, & World Health a/%20file/429125/future-cities-globalagenda.
Organization. The Selection and Use of Essential pdf
Medicines: Report of the WHO Expert (accessed on 1 January 2020)
Commiee, 2013 (Including the 18th WHO [42] Eremia, M.; Toma, L.; Sanduleac, M. The
Model List of Essential Medicines and the 4th smart city concept in the 21st century. Procedia
WHO Model List of Essential Medicines for Eng. 2017, 181, 12–19.
Children); World Health Organization: Geneva, [43] Komninos, N. Intelligent Cities: Innovation,
Switzerland, 2014; Volume 985. Knowledge Systems and Digital Spaces;
[33] Washburn, D.; Sindhu, U.; Balaouras, S.; Routledge: Oxford, UK, 2013.
Dines, R.A.; Hayes, N.; Nelson, L.E. Helping CIOs [44] Allwinkle, S.; Cruickshank, P. Creating
Understand “Smart City” Initiatives. Growth Smart-er Cities: An Overview. J. Urban Technol.
2010, 17, 1–17. 2011, 18, 1–16.
[34] Mills, D.; Pudney, S.; Pevcin, P.; Dvorak, J. [45] Kitchin, R. Making sense of smart cities:
Addressing present shortcomings. Camb. J. Reg. R.; Jara, A.J.; Skarmeta, A.G. A Global
Econ. Soc. 2015, 8, 131–136. Perspective of Smart Cities: A Survey. In
[46] Kitchin, R.; Lauriault, T.P.; McArdle, G. Proceedings of the 2013 Seventh International
Knowing and governing cities through urban Conference on Innovative Mobile and Internet
indicators, city benchmarking and real-time Services in Ubiquitous Computing, Taichung,
dashboards. Reg. Stud. Reg. Sci. 2015, 2, 6–28. Taiwan, 3–5 July 2013; pp. 439–444.
[47] Manville, C.; Europe, R.; Millard, J.;
Institute, D.T.; Liebe, A. Mapping Smart Cities in 25. 1. Kim, S.T. A review and suggestion on the
the EU. 2014. Available online: methodology of domestic content analysis
hps://www.europarl.europa.eu/RegData/etud research. Commun. Theory 2005, 1, 39–67.
es/etudes/join/2014/507480/IPOLITRE_ 26. 2. Chung, S.; Yu, P.J. Analysis on E-
ET(2014)507480_EN.pdf Participation Research Trend in Korea.
(accessed on 1 January 2020). Informatization Policy 2014, 21, 3–23. [CrossRef]
[48]. Giffinger, R.; Fertner, C.; Kramar, H.; 27. 3. Kim, G.J. Meta-analysis of research on the
Meijers, E. City-ranking of European Medium- spread of innovation: Focusing on the field of
Sized Cities. Cent. Reg. Sci. Vienna UT 2007, 9, journalism. Korea J. 2010, 54, 31–56.
13. 28. 4. Mok, J.-W.; Kang, K.-B.; Oh, C.-H.; Choi, Y.;
[49]. Chourabi, H.; Nam, T.; Walker, S.; Gil- Hong, H.; Kim, H.; Park, Y. An Empirical Study on
Garcia, J.R.; Mellouli, S.; Nahon, K.; Pardo, T.A.; the Trends of Korean Policy Research: Focused
Scholl, H.J. Understanding smart cities: An on Lasswell’s Policy Paradigm. Korean Policy
integrative framework. In Proceedings of the Stud. Rev. 2005, 14, 131–154.
2012 45th Hawaii International Conference on 29. 5. Kim, K. Understanding the Social Science
System Sciences, Maui, HI, USA, 4–7 January Research Methodology; B&M Books: Seoul,
2012; IEEE: Piscataway, NJ, USA, 2012; pp. Korea, 2011.
2289–2297. 30. 6. Park, K.C.; Lee, C.H. A study on the
[50]. Joshi, S.; Saxena, S.; Godbole, T.; Shreya. research trends for smart city using topic
Developing Smart Cities: An Integrated modeling. J. Internet Comput. Serv. 2019, 20,
Framework. Procedia Comput. Sci. 2016, 93, 119–128
902–909.
[51]. Anand, A.; Winfred Rufuss, D.D.; Rajkumar, 67. 7. Taylor, L.; Nitschke, G. Improving Deep
V.; Suganthi, L. Evaluation of Sustainability Learning with Generic Data Augmentation. In
Indicators in Smart Cities for India Using MCDM Proceedings of the 2018 IEEE Symposium
Series on Computational Intelligence (SSCI),
Approach. Energy Procedia 2017, 141, 211–215. Bangalore, India, 18–21 November 2018;
[CrossRef] Volume 1708, pp. 1542–1547
[52]. Zadeh, L.A. Fuzzy logic. Computer 1998,
21, 83–93. 116 8. Abdulnabi, A.H.; Wang, G.; Lu, J.; Jia, K.
[53]. Zadeh, L.A. Is there a need for fuzzy logic? Multi-Task CNN Model for Attribute
Inf. Sci. 2008, 178, 2751–2779. Prediction. IEEE Trans. Multimedia 2015, 17,
[54]. Dzitac, I.; Filip, F.G.; Manolescu, M.-J. 1949–1959
Fuzzy
117 9. Perez, L.; Wang, J. The Effectiveness of
Logic Is Not Fuzzy: World-renowned Computer Data Augmentation in Image Classification using
Scientist Loi A. Zadeh. Int. J. Comput. Deep Learning. arXiv 2017, arXiv:1712.04621.
Commun. Control. 2017, 12, 748–789. [Google Scholar]
[55]. Neiro, P.; De Marco, A.; Cagliano, A.C.; 118 10. Chen, Y.; Lin, Z.; Zhao, X.; Wang, G.; Gu,
Mangano, G.; Scorrano, F. Current trends in Y. Deep Learning-Based Classification of
Hyperspectral Data. IEEE J. Sel. Top. Appl. Earth
Smart City initiatives: Some stylised facts. Cities
Obs. Remote Sens. 2014, 7, 2094–2107.
2014, 38, 25–36. [Google Scholar] [CrossRef]
[56]. Pellicer, S.; Santa, G.; Bleda, A.L.; Maestre,
14. 11. Wirth, N. Hello marketing, what can artificial 29. 26. Yeung, K. “Hypernudge”: Big Data as a mode of
intelligence help you with? Int. J. Mark. Res. 2018, 60, regulation by design. Inf. Commun. Soc. 2017, 20, 118–
435–438. [CrossRef] 136. [CrossRef]
15. 12. Cugurullo, F. Frankenstein Urbanism: Eco, Smart 30. 27. Ranchords, S. Nudging citizens through technology
and Autonomous Cities, Artificial Intelligence and the End in smart cities. Int. Rev. Law Comput. Technol. 2020, 34,
of the City, 1st ed.; Routledge: 254–276. [CrossRef]
Abingdon, UK; New York, NY, USA, 2021. [CrossRef]
16. 13. Lloyd, D. Frankenstein’s children: Artificial
intelligence and human value. Metaphilosophy 1985, 16,
307–318. [CrossRef]
17. 14. Grossi, G.; Pianezzi, D. Smart cities: Utopia or
neoliberal ideology? Cities 2017, 69, 79–85. [CrossRef]
18. 15. Albino, V.; Berardi, U.; Dangelico, R.M. Smart Cities:
Definitions, Dimensions, Performance, and Initiatives. J.
Urban Technol.
2015, 22, 3–21. [CrossRef]
19. 16. Burns, R.; Fast, V.; Levenda, A.; Miller, B. Smart
cities: Between worlding and provincialising. Urban Stud.
2021, 58, 461–470.
[CrossRef]
20. 17. Almirall, E.; Wareham, J.; Ratti, C.; Conesa, P.; Bria,
F.; Gaviria, A.; Edmondson, A. Smart Cities at the
Crossroads: New Tensions
in City Transformation. Calif. Manag. Rev. 2016, 59, 141–
152. [CrossRef]
21. 18. Park, S.; Lee, S.; Park, S.; Park, S. AI-Based Physical
and Virtual Platform with 5-Layered Architecture for
Sustainable Smart
Energy City Development. Sustainability 2019, 11, 4479.
[CrossRef]
22. 19. Lazaroiu, G.C.; Roscia, M. Definition methodology
for the smart cities model. Energy 2012, 47, 326–332.
[CrossRef]
23. 20. Ahvenniemi, H.; Huovila, A.; Pinto-Seppä, I.;
Airaksinen, M. What are the differences between
sustainable and smart cities?
Cities 2017, 60, 234–245. [CrossRef]
24. 21. Kramers, A.; Höjer, M.; Lövehagen, N.;Wangel, J.
Smart sustainable cities–Exploring ICT solutions for
reduced energy use in
cities. Environ. Model. Softw. 2014, 56, 52–62. [CrossRef]
25. 22. Hoang, A.T.; Pham, V.V.; Nguyen, X.P. Integrating
renewable sources into energy system for smart city as a
sagacious strategy
towards clean and sustainable process. J. Clean. Prod.
2021, 305, 127161. [CrossRef]
26. 23. Adhikari, M.; Ghimire, L.P.; Kim, Y.; Aryal, P.;
Khadka, S.B. Identification and Analysis of Barriers against
Electric Vehicle Use.
Sustainability 2020, 12, 4850. [CrossRef]
27. 24. Sanguesa, J.A.; Torres-Sanz, V.; Garrido, P.;
Martinez, F.J.; Marquez-Barja, J.M. A Review on Electric
Vehicles: Technologies and
Challenges. Smart Cities 2021, 4, 372–404. [CrossRef]
28. 25. Omitaomu, O.A.; Niu, H. Artificial Intelligence
Techniques in Smart Grid: A Survey. Smart Cities 2021, 4,
548–568. [CrossRef]