Submitted:
07 July 2026
Posted:
08 July 2026
You are already at the latest version
Abstract
The sustained growth of urban areas has increased the complexity of managing services, infrastructure, and mobility, creating a need for advanced technological solutions capable of responding dynamically to rapidly changing environments. In this context, adaptive smart urban systems have emerged as an innovative alternative that integrates artificial intelligence (AI) to optimize real-time decision-making. This study presents a systematic literature review focused on the frameworks underpinning these systems, examining the main adaptive AI techniques, current technological trends, and the structural challenges that limit their implementation. The methodology applied is based on the selection and critical analysis of indexed scientific publications, enabling the identification of predominant approaches such as machine learning, deep learning, multi-agent systems, and reinforcement learning. The findings reveal a strong convergence between AI, the Internet of Things (IoT), and Big Data, as well as significant limitations in terms of interoperability, data governance, and scalability. It is concluded that, while the advances are promising, the consolidation of these systems requires a comprehensive approach that combines technological innovation, appropriate regulation, and social sustainability.
Keywords:
smart cities
; adaptive artificial intelligence
; urban systems
; machine learning
; smart infrastructure
1. Introduction
Contemporary cities face growing challenges associated with digital transformation and the accelerated process of urbanization, which has emerged as one of the most significant phenomena of the 21st century. Population growth, territorial expansion, and the rising demand for urban services have created highly complex environments, characterized by shifting dynamics and strong interdependence among systems. This scenario calls for the development of technological solutions that are not only efficient and scalable but also capable of continuously adapting to changing conditions and uncertain contexts.
In this context, the concept of adaptive smart urban systems has emerged; these systems are distinguished by their ability to integrate multiple data sources, analyze them in real time, and autonomously adjust their behavior. These systems represent a significant evolution from traditional approaches to urban management, as they incorporate continuous learning mechanisms that allow them to progressively improve their performance across different operational scenarios.
Artificial intelligence plays a fundamental role in these types of systems by providing advanced tools for pattern modeling, behavior prediction, and the optimization of complex processes. Unlike conventional approaches, which are based on static rules, AI-driven adaptive models enable the creation of dynamic systems that evolve based on data and the environment, facilitating more accurate and context-aware decision-making [1,2,3,4].
However, the implementation of these solutions faces multiple limitations from both a technological and organizational perspective. Among these are limited interoperability between platforms, the difficulty of integrating with existing infrastructure, and the heterogeneity of data sources. Added to this are regulatory and ethical considerations, such as privacy protection, algorithmic transparency, and urban data governance, which significantly influence their adoption [5,6].
Smart urban systems are characterized by the convergence of various emerging technologies, including the Internet of Things (IoT), cloud computing, big data analytics, and edge computing. These technologies enable the capture, processing, and analysis of information in real time, facilitating the management of urban domains such as transportation, energy, security, and governance. Consequently, they promote the development of more efficient, sustainable, and resilient urban environments [3,4,7,8].
Within this framework, the purpose of this article is to analyze, through a systematic literature review, the main frameworks used in the design and implementation of adaptive smart urban systems. In particular, it seeks to identify the predominant artificial intelligence techniques, examine emerging technological trends, and analyze the structural challenges that influence their adoption in different urban contexts.
To this end, the methodology is based on the collection, selection, and critical analysis of scientific articles indexed in recognized academic databases, which allows for the identification of predominant approaches such as machine learning, deep learning, agent-based systems, and hybrid artificial intelligence models.
Despite the progress made, the implementation of adaptive smart urban systems continues to face significant structural challenges. Among these are scalability issues, limited interoperability between heterogeneous systems, the complexity of data management and governance, as well as the economic and technological constraints specific to each urban context. These limitations highlight the need to critically analyze existing frameworks, evaluating their capabilities, limitations, and potential for adaptation [5,8,9,10].
In this regard, this study proposes a systematic review aimed at identifying, classifying, and analyzing the main frameworks used in the development of adaptive smart urban systems, as well as at establishing a comprehensive overview of emerging trends and the structural guidelines necessary for their effective implementation, thereby contributing to the advancement of knowledge in the field of smart cities.
2. Previous Work
Over the past decade, research on smart urban systems has experienced sustained growth, driven by the increase in the urban population and the growing complexity of public service management. Various studies agree that traditional models are insufficient to address issues such as congestion, energy consumption, and environmental sustainability. In this context, the integration of digital technologies and the intensive use of data have emerged as fundamental pillars of urban transformation [3,7,11,12].
The concept of the smart city has evolved from approaches focused on technological infrastructure toward more holistic models that incorporate governance, sustainability, and citizen participation. In this sense, a smart city is defined not only by the availability of technology, but by its ability to transform data into efficient and adaptive decisions [13,14,15]. This paradigm shift has driven the conception of urban systems as dynamic and interconnected ecosystems.
In this context, artificial intelligence has established itself as one of the primary tools for urban management. Various studies demonstrate its usefulness in predicting, optimizing, and automating services, particularly in areas such as transportation, energy, and security [1,8,9,16,17,18]. However, although these approaches have demonstrated high effectiveness, many of them have limitations in terms of adaptability to highly dynamic environments.
Regarding the techniques employed, supervised machine learning continues to be widely used due to its predictive capabilities in structured scenarios. However, its reliance on labeled data limits its application in complex urban contexts where information is heterogeneous and ever-changing [19,20,21].
Unsupervised learning has made it possible to identify hidden patterns in large volumes of urban data, facilitating spatial segmentation and behavioral analysis. Despite their exploratory utility, these methods often lack interpretability and control in decision-making [22,23,24,25,26].
Deep learning has expanded analytical capabilities in the processing of complex data, particularly in computer vision and temporal analysis applications. However, its high computational cost and low explainability pose significant challenges for its implementation in real-world urban environments [27,28,29,30].
In terms of infrastructure, architectural frameworks integrate technologies such as IoT, cloud computing, and data analytics to enable the continuous capture and processing of urban information. While these models allow for scalability, they have limitations related to latency and dependence on connectivity [31,32,33,34].
As an alternative, edge computing has emerged as a solution that allows data to be processed close to its source, improving response times and reducing the load on centralized systems. However, its implementation poses challenges in terms of coordination and resource allocation [35,36,37].
An emerging trend is the development of adaptive urban systems capable of adjusting their behavior in real time through continuous learning mechanisms. These systems represent an evolution beyond static models; however, their implementation still faces limitations related to technological integration and the availability of real-time data [4,38,39].
In the field of urban mobility, numerous studies have focused on optimizing traffic flow through predictive models and intelligent control systems. Although these approaches have demonstrated improvements in transportation efficiency, their performance can be affected by the variability of the urban environment [4,27,40].
Significant challenges remain, such as a lack of interoperability, the absence of common standards, and the difficulty of integrating heterogeneous systems [5,41,42]. Furthermore, economic factors and operational constraints hinder the scalability of these solutions in real-world contexts [9,43,44].
Finally, issues such as privacy, cybersecurity, and transparency in decision-making have become increasingly important, driving the development of approaches based on explainable artificial intelligence and data governance [45,46,47].
In summary, while the literature shows significant progress in the development of smart urban systems, there are still major limitations in terms of adaptability, integration, and scalability. These gaps justify the need for a systematic review aimed at identifying more robust and adaptive approaches to managing complex urban environments.
3. Methodology
3.1. Research Approach
This research is conducted using a qualitative approach of a descriptive and analytical nature, based on a systematic literature review. This approach allows for a structured examination of existing knowledge regarding adaptive smart urban systems, facilitating the identification of trends, predominant artificial intelligence techniques, and the main challenges associated with their implementation.
Unlike purely quantitative studies, this approach prioritizes the critical interpretation of scientific evidence, allowing us not only to describe the state of the art but also to identify patterns, relationships, and research gaps. Furthermore, it enables comparisons between different frameworks and architectures proposed in the field of smart cities, contributing to a more comprehensive understanding of the phenomenon under study.
3.2. Search Strategy
The search strategy was designed to ensure the collection of relevant, up-to-date, and methodologically sound literature. To this end, the Scopus database was selected due to its broad coverage of indexed publications and its recognition within the international scientific community.
The search query was constructed by combining key terms related to the subject of study, such as smart cities, adaptive artificial intelligence, urban intelligent systems, machine learning in urban environments, and IoT smart city frameworks. These terms were combined using Boolean operators (AND, OR), which allowed the results to be broadened or narrowed based on their relevance.
Additionally, time filters were applied covering the period 2015–2026, in order to prioritize recent research and reflect the current state of knowledge in the field. Furthermore, the results were limited to scientific articles and conference proceedings in English, thereby ensuring the quality and relevance of the selected sources.
Table 1.
Search strategy used in the systematic review
| Element | Description |
|---|---|
| Database | Scopus |
| Keywords | smart cities, adaptive artificial intelligence, urban intelligent systems, machine learning in urban environments, IoT smart city frameworks |
| Operators | AND, OR |
| Time period | 2015 – 2026 |
| Document Type | Articles and conference papers |
| Language | English |
| Initial Results | 2,528 records |
3.3. Study Selection Process
The study selection process was conducted in accordance with the PRISMA guidelines, which ensured the transparency, traceability, and reproducibility of the methodological procedure.
During the identification phase, a total of 2,528 records were retrieved using the defined search strategy. Subsequently, a screening phase was conducted based on a review of titles, abstracts, and keywords, during which duplicate records, studies unrelated to the urban context, and those that did not incorporate artificial intelligence techniques were eliminated. As a result of this initial screening, a subset of 105 potentially relevant articles was obtained.
During the eligibility phase, a thorough review of the full text of the selected studies was conducted. At this stage, criteria such as thematic relevance, methodological quality, scientific contribution, and applicability to smart urban systems were evaluated. Following this analysis, 64 articles were finally selected, constituting the study’s final sample.
3.4. Inclusion and Exclusion Criteria
To ensure the consistency and objectivity of the selection process, inclusion and exclusion criteria were defined in advance. Only publications from 2015 to 2026, indexed in recognized databases and focused on the study of smart cities or urban systems, were considered. Additionally, research incorporating artificial intelligence techniques and proposing models, architectures, or frameworks was included.
On the other hand, duplicate studies, works without application in urban environments, publications not focused on artificial intelligence, documents without full-text access, and those that did not meet minimum standards of methodological rigor were excluded.
3.5. Materials and Methods
The final sample consisted of 64 scientific articles, which provide an up-to-date overview of the state of the art in adaptive smart urban systems. The analysis was conducted by integrating a systematic review, a comparative analysis, and a thematic classification.
The studies were organized based on the artificial intelligence techniques used, the application domains—such as transportation, energy, or security—and the type of architecture proposed. This organization allowed us to structure the information in a coherent manner and facilitated its subsequent analysis.
3.6. Data Structure
For each selected study, a systematic process of extracting relevant information was carried out. The main attributes analyzed included the name of the framework, the technologies used, the type of architecture, the application context, the region of implementation, the year of publication, and the associated technological standards.
This structuring enabled the creation of a comparative database, which served as input for subsequent analysis.
3.7. Data Analysis
The analysis of the information was conducted using a qualitative, interpretive, and comparative approach. In this regard, the studies were classified by theme, the predominant techniques—such as machine learning, deep learning, multi-agent systems, and reinforcement learning—were identified, and the main emerging technological trends were analyzed.
In addition, the limitations and challenges reported in the literature were evaluated, providing a comprehensive overview of the current state of research.
3.8. Identification of Problems
Based on the analysis of the selected studies, various structural problems affecting the implementation of adaptive smart urban systems were identified. Among the main challenges are limited interoperability between heterogeneous systems, scalability issues, dependence on high-quality data, and the complexity of integrating multiple technologies.
These challenges are associated with factors such as the lack of common technological standards, the coexistence of heterogeneous infrastructures, budgetary constraints, and technological gaps between different regions.
3.9. Classification of Frameworks
The studies were classified according to their predominant technological approach, identifying architectures based on the Internet of Things, adaptive systems driven by artificial intelligence, solutions supported by edge computing, and models oriented toward urban sustainability.
The application of the PRISMA model, together with the qualitative analysis conducted, ensures the validity, reliability, and reproducibility of the study. The results obtained form the basis for the analysis presented in the following section.
3.10. Bibliometric Analysis Derived from the Systematic Review
Based on the results obtained from the systematic literature review, a bibliometric analysis was conducted with the aim of identifying the main research trends, emerging topics, the temporal evolution of scientific output, and the conceptual relationships among the most representative keywords in the field. This analysis complemented the qualitative selection of studies with a quantitative perspective, facilitating an understanding of the academic development of adaptive smart urban systems and the identification of opportunities for future research.
4. Results
4.1. Results of the Selection Process
The systematic review process identified a total of 2,528 initial records, obtained using the search strategy defined in the Scopus database. Subsequently, through an initial screening process based on a review of titles, abstracts, and keywords, duplicates and studies unrelated to the urban context or not involving the application of artificial intelligence were eliminated, reducing the sample to 105 potentially relevant articles.
In a second phase, corresponding to the eligibility analysis, the selected documents were read in full, and their thematic relevance, methodological rigor, and scientific contribution were evaluated. As a result, studies that did not meet the established criteria were excluded, ultimately yielding a sample of 64 scientific articles that form the basis of the analysis.
This process is summarized in Table 2, which provides a structured overview of the progressive reduction of studies through the phases of the PRISMA model.
4.2. Bibliometric Analysis of the Dataset
4.2.1. General Description
The analyzed dataset consists of the 64 selected articles, which represent a significant sample of the current state of research on adaptive smart urban systems. These studies cover the period from 2012 to 2026, demonstrating a progressive evolution of the field.
Overall, there is a high number of authors per paper, reflecting a strong collaborative component in the research. Furthermore, the presence of international co-authorship indicates that the development of solutions for smart cities is an area of global interest.
4.2.2. Annual Scientific Output
An analysis of scientific output reveals an upward trend in the number of publications, particularly in recent years. This growth is directly related to advances in technologies such as artificial intelligence, the Internet of Things (IoT), and urban data analytics.
4.2.3. Analysis of Sources and Authors
An analysis of the sources reveals that most of the studies come from high-impact indexed journals and international conferences, which guarantees the scientific quality of the selected works.
As for the authors, a diverse distribution is observed, with no dominant concentration, indicating that the field is expanding and has multiple active contributors. This diversity fosters the development of multidisciplinary approaches to creating smart urban solutions.
4.2.4. Keyword Frequency
The keyword analysis identified the most representative terms in the literature. Among the most frequent are:
- Smart city
- Artificial intelligence
- Internet of things
- Machine learning
- Sustainability
These terms clearly reflect the technological and data-driven approach that characterizes modern smart urban systems.
4.2.5. Keyword Co-occurrence Network
The co-occurrence network shows the relationships between the main concepts used in the literature. It reveals that the term “smart city” acts as a central node, establishing strong connections with artificial intelligence, IoT, and sustainability.
4.2.6. Thematic Grouping
Cluster analysis made it possible to identify different lines of research within the domain studied. Among the main thematic groups are:
- IoT-based infrastructure
- AI-driven analytics
- Sustainability-oriented systems
- Urban governance and management
4.2.7. Emerging Trends
Trend analysis reveals a growing interest in advanced technologies such as:
- Explainable artificial intelligence
- Edge computing
- Smart energy systems
4.2.8. Geographic Distribution of Research
An analysis of the selected studies shows that scientific output is concentrated primarily in regions such as Europe, Asia, and North America.
This suggests that countries with greater technological development are leading the way in smart city research, while other regions have a lower level of participation, which opens up opportunities for future research in local contexts.
Figure 5.
Geographic Distribution of Research
Overall, the results of the bibliometric analysis show that research on adaptive smart urban systems is in a phase of growth and consolidation. The combination of artificial intelligence, the Internet of Things (IoT), and distributed architectures has enabled the development of increasingly sophisticated solutions for urban management.
Persistent challenges related to technological integration, interoperability between systems, and the need for sustainable and scalable approaches have also been identified, which is consistent with the findings of recent studies [3,8,41].
These findings form the basis for the comparative analysis of the frameworks, which is presented in the following sections.
4.3. Classification of Frameworks for Smart Urban Systems
Based on a detailed analysis of the 64 selected articles, different approaches to the design and implementation of frameworks for smart urban systems were identified. These approaches were grouped according to their technological orientation, architecture, and functional purpose, allowing for the establishment of a structured classification that facilitates an understanding of the current state of the field.
The proposed classification is organized into four main categories:
- Internet of Things (IoT)-based architectures
- Artificial intelligence-driven frameworks
- Distributed architectures (edge/fog computing)
- Approaches focused on sustainability and governance
4.3.1. Relevant Research on Internet of Things (IoT)-Based Architectures
IoT-based frameworks form the technological foundation of most smart urban systems. These approaches are characterized by the integration of sensor devices, communication networks, and data collection platforms that enable real-time monitoring of various variables in the urban environment. The reviewed studies show an evolution from traditional IoT architectures—focused on data capture and transmission—toward hybrid approaches that incorporate artificial intelligence, digital twins, Zero Trust security, and intelligent cyber-physical systems.
In these models, the architecture is typically structured in multiple layers, including perception (sensors), communication (networks), processing (cloud or edge platforms), and application (urban services). This organization facilitates the continuous acquisition of data and its subsequent analysis for decision-making. Table 3 summarizes the main frameworks identified within this category, listing their name, authors, year, country or context of presentation, and main contribution.
An analysis of these frameworks shows that the role of the IoT is no longer limited to urban sensor deployment. In the most recent proposals, the data captured by sensors becomes input for predictive models, smart maintenance systems, adaptive mobility platforms, and security mechanisms for model training. In this regard, the hybrid AI-IoT framework with digital twin integration represents a clear transition toward AIoT architectures, where urban infrastructure can be simulated and evaluated before actual failures occur [7]. Complementarily, IoT applied to traffic gains value when combined with algorithms capable of processing data in real time and generating optimization actions [4].
TwinSnake broadens the discussion to the security of AIoT ecosystems, emphasizing that urban adaptability requires protecting the data used to train intelligent models [37]. Intelligent cyber-physical systems underscore the importance of connecting physical objects, sensors, and computational processing to achieve continuous interaction between the city and its digital platforms [31]. Finally, the AI-IoT convergence can be interpreted as a structural trend toward a new generation of urban systems that are more autonomous and context-aware [32]. Taken together, these frameworks show that the main strength of IoT architectures is their ability to generate real-time urban data; however, their limitations continue to relate to interoperability, security, data quality, and dependence on robust connectivity infrastructures.
4.3.2. Relevant Research on Artificial Intelligence
A second category consists of frameworks focused on using artificial intelligence as the core of the system. These approaches incorporate machine learning algorithms, deep learning, explainable artificial intelligence, and multi-agent systems to analyze urban data and generate adaptive responses. Unlike traditional models, these systems not only collect information but are also capable of predicting urban events, optimizing resources, automating decisions, and dynamically adapting to changes in the environment.
Applications within this category include public services, smart transportation, urban infrastructure, land-use planning, and algorithmic transparency. Table 4 presents the main AI-driven frameworks identified in the review, listing their name, authors, year, country or context, and main contribution.
An analysis of these frameworks shows that artificial intelligence is used as a layer of reasoning capable of transforming large volumes of urban data into actionable knowledge. SmartCityPredict and SmartCityNext reflect this approach by employing machine learning and real-time analytics to improve the efficiency and responsiveness of urban services [16,21]. In the field of mobility, AI-based smart transportation architecture demonstrates a more concrete application, tied to a specific geographic context, where algorithms support traffic management and real-time decision-making [20].
AI-driven decision-making architectures extend the scope of these systems to urban infrastructures prepared for future scenarios, in which automation enables a response to complex operational conditions [8]. However, the growing autonomy of these models requires mechanisms for transparency and trust; therefore, explainable artificial intelligence becomes a key component for understanding the reasoning behind the decisions generated by algorithms [45]. Finally, multi-agent approaches provide a distributed perspective for urban planning by allowing multiple intelligent agents to interact, evaluate scenarios, and support more flexible decision-making processes [38].Taken together, these results show that AI has evolved from applications focused solely on prediction to more complex architectures geared toward decision-making, algorithmic transparency, and the automation of urban processes.
4.3.3. Relevant Work on Distributed Architectures: Edge and Fog Computing
The growth of real-time applications has driven the development of distributed architectures, particularly those based on edge computing, fog computing, Edge AI, federated learning, and 5G networks. These frameworks aim to reduce latency and improve processing efficiency by executing analytical tasks close to the data source.
Unlike centralized cloud models, these architectures enable local data processing, real-time response, reduced network traffic, and greater system resilience. Table 5 summarizes the main distributed frameworks identified in the review.
An analysis of these frameworks reveals a growing trend toward decentralized architectures that seek to minimize latency, improve citizens’ privacy, and increase the resilience of smart urban systems. In this regard, proposals based on edge computing and edge intelligence reduce dependence on central servers by processing information on devices close to the source of data generation [35,36]. Federated learning, in turn, extends this logic by enabling the collaborative training of models without moving the original data, which promotes scalability and cooperation among multiple urban stakeholders [33].
The incorporation of 5G and multi-agent reinforcement learning accelerates the response to dynamic events, especially in urban infrastructures that require low latency and continuous adaptation [29]. The integration of blockchain adds an additional layer of traceability and trust, which is particularly relevant when distributed urban systems use sensitive data or AI models shared among different entities [49]. This approach is especially useful in critical applications such as traffic control, security monitoring, and emergency management; however, it introduces new challenges related to distributed management, data synchronization, and the maintenance of multiple nodes.
4.3.4. Relevant Works on Sustainability- and Governance-Oriented Approaches
A set of frameworks was identified that integrate technological aspects with social, environmental, and governance dimensions. These approaches seek not only to optimize the city’s operations but also to ensure its long-term sustainability. This category includes models that incorporate energy efficiency, emissions reduction, citizen participation, transparency in decision-making, ethical data management, climate resilience, and public dissemination of urban information.
Table 6 presents the main sustainability- and governance-oriented frameworks identified in the review.
An analysis of these frameworks reveals a gradual transition from approaches focused exclusively on technology toward comprehensive models that integrate sustainability, data governance, and urban resilience. Frameworks based on digital twins enable the dynamic representation of urban infrastructure and support both operational management and the public communication of information [1]. In the case of transportation, the combination of digital twins and real-time optimization helps reduce environmental impacts and improve the efficiency of mobility networks [39].
Data governance emerges as a cross-cutting theme, as it defines the conditions under which urban data can be collected, shared, audited, and used ethically. In this regard, the governance models evaluated for European smart cities provide relevant regulatory and operational criteria [5], while SORA-ATMAS extends this discussion to adaptive trust management and the alignment of systems based on multiple LLMs [50]. Furthermore, approaches focused on climate change and sustainability show that the convergence of AIoT and AI can strengthen urban resilience, optimize resources, and promote more sustainable urban lifestyles [9,43,44].These frameworks reflect an evolution of the smart city concept toward a more comprehensive approach, in which technology serves as a means to improve quality of life, environmental sustainability, and institutional trust.
4.3.5. Summary of the Classification
In summary, the proposed classification shows that smart city frameworks have evolved from isolated technological structures toward complex, adaptive, and multidimensional ecosystems.
The results obtained lead to the conclusion that the current trend is toward the integration of multiple technologies, automation based on artificial intelligence, distributed processing, and the incorporation of sustainability criteria.
This classification provides a solid conceptual foundation for the development of new models of adaptive smart urban systems, as well as for identifying opportunities for future research.
4.4. Identified and Implemented Frameworks
4.4.1. Artificial Intelligence-Driven Frameworks
Within this category, one framework was identified that has an application associated with a specific geographic context. These types of architectures are characterized by their use of artificial intelligence algorithms as a central element for analyzing urban information, optimizing processes, and generating adaptive responses to changes in the environment.
Project: An AI-Driven Intelligent Transportation System: Functional Architecture and Implementation
- Framework implemented: CityAI.
- Implementation: Applied in an urban context in Hungary, Europe, as a smart architecture geared toward the management and optimization of Intelligent Transportation Systems (ITS).
The CityAI framework, developed by Huszak, Simon, Bokor, Tizedes, and Pekar in 2024, integrates technologies such as artificial intelligence, machine learning, functional architectures for urban transportation management, and data processing mechanisms geared toward decision-making.
The proposed architecture enables the collection, processing, and interpretation of information related to urban mobility, using computational models capable of analyzing the behavior of the transportation system and generating useful insights to improve its operation. Through the application of intelligent techniques, the framework seeks to adapt to changes in the urban environment and provide data-driven solutions to optimize traffic-related processes.
One of the most significant aspects of this framework is its implementation in a specific geographic setting, which demonstrates the applicability of artificial intelligence within real-world mobility systems. Unlike purely conceptual proposals, CityAI presents a functional architecture in which intelligent models are integrated with transportation management to transform urban data into useful information for vehicle planning and operations.
This framework represents a significant contribution to the development of solutions focused on analyzing mobility patterns, optimizing traffic, and enabling smart urban transportation planning. Its implementation highlights the importance of combining artificial intelligence, data processing, and digital infrastructure to address the current challenges facing smart cities.
4.4.2. Approaches Focused on Sustainability and Governance
Within this category, a framework focused on smart urban information management was identified, in which digital technologies are used as tools to strengthen sustainability, governance, and decision-making within urban environments. These approaches are characterized by the integration of different data sources, digital platforms, and information management mechanisms with the aim of improving the operational efficiency and transparency of public services.
Unlike frameworks focused exclusively on data processing or the optimization of specific processes, these models incorporate a comprehensive perspective in which technology serves as a support for urban planning, efficient resource management, and the generation of data-driven knowledge.
Project: Smart City Digital Twin Framework for Real-Time Multi-Data Integration and Wide Public Distribution
- Framework implemented: Snap4City.
- Implementation: Applied in Italy as a smart city-oriented architecture for the integration, management, and distribution of urban information using digital technologies.
The Snap4City framework, developed by Adreani, Bellini, Fanfani, Nesi, and Pantaleo in 2024, is an open-source, integrated IoT platform architecture designed to manage the entire urban data lifecycle. The framework enables the collection, indexing, processing, and distribution of information from multiple sources, facilitating the creation of dynamic digital representations of the urban environment.
The architecture integrates technologies such as digital twins (Digital Twin), the Internet of Things (IoT), smart urban data management, and digital platforms for the public distribution of information. Through this integration, the framework enables the centralization of heterogeneous data generated by different urban systems, providing tools for the visualization, analysis, and monitoring of variables relevant to city management.
A key feature of Snap4City is its ability to combine information from various systems within a single platform, thereby improving interoperability among urban services and facilitating data-driven decision-making processes. Its implementation in Italy demonstrates the practical application of digital architectures geared toward the development of smart cities and the efficient management of urban resources.
This framework represents a significant contribution to sustainability and governance approaches because its purpose is not limited to the technological processing of information; rather, it seeks to strengthen urban administration through mechanisms of integration, transparency, and access to data. Although it is not directly focused on identifying traffic congestion using artificial intelligence, it provides a relevant technological infrastructure for future smart mobility systems that require the integration of data from sensors, connected devices, and urban platforms.
Project: Green Transportation Planning for Smart Cities: Digital Twins and Real-Time Traffic Optimization in Urban Mobility Networks
- Implemented framework: Digital twins (DT) with real-time traffic optimization systems.
- Implementation: Applied in Poland within the context of sustainable urban transportation planning and mobility network optimization.
The framework developed by Lis and Madziel in 2026 proposes an architecture based on the integration of digital twins, microsimulation, and real-time information processing to improve urban transportation management. The proposal combines digital models capable of representing the actual conditions of the mobility system with analytical mechanisms designed to evaluate scenarios and optimize the operation of road networks.
The architecture uses a dynamic Adaptive Flow Measurement (AIM) logic, which interacts with data from IoT sensors to enable real-time Software-in-the-Loop (SiL) simulation processes. This integration makes it possible to analyze traffic behavior, identify potential congestion scenarios, and evaluate optimization strategies before implementing them in the physical environment.
The framework integrates technologies such as digital twins (Digital Twin), Intelligent Transportation Systems (ITS), real-time traffic optimization, digital urban modeling, and sustainable mobility systems. Through this combination of technologies, the model generates a dynamic representation of the transportation system, facilitating data-driven decision-making and improving the ability to respond to changes in traffic conditions.
A key feature of this framework is its preventive approach to traffic congestion, as it uses simulations and digital models to anticipate system behavior before implementing changes to infrastructure or mobility strategies. This makes it possible to evaluate alternatives aimed at increasing operational efficiency, reducing environmental impacts, and improving the sustainability of urban transportation.
This framework represents a significant contribution to approaches focused on sustainability and governance, as it combines advanced digital technologies with objectives related to energy efficiency, resource optimization, and smart mobility planning.Although it does not focus directly on GPS trajectory clustering algorithms, its architecture is relevant to intelligent transportation systems that require the integration of urban data, computational models, and real-time analysis for traffic congestion management.
4.4.3. Comparison of the Identified Frameworks
A comparison of the identified frameworks shows that intelligent systems applied to urban environments do not rely on a single technology, but rather are evolving toward integrated architectures capable of combining different approaches to address the complexity of modern cities. Each framework analyzed has specific characteristics depending on its primary objective, the technologies used, and its level of integration with urban transportation management.
Table 7.
Comparison of the Identified and Implemented Frameworks
| Framework | Approach | Main Technologies | Contribution to Smart Transportation |
|---|---|---|---|
| CityAI | Artificial Intelligence applied to Intelligent Transportation Systems (ITS) | Artificial Intelligence, Machine Learning, data processing, and functional architectures for urban transportation | Enables the analysis of mobility information, the generation of insights from urban data, and the improvement of operational transportation management through intelligent models. |
| Snap4City | Intelligent urban data management and governance | Digital Twin, Internet of Things (IoT), integration of multiple data sources, and digital urban information platforms | Facilitates the integration, processing, and distribution of urban data, providing a technological infrastructure for the efficient administration of smart cities. |
| Digital Twin (DT)-Based Framework for Real-Time Traffic Optimization | Traffic Optimization and Sustainable Mobility | Digital Twins, IoT, Microsimulation, Adaptive Flow Measurement (AIM), and Software-in-the-Loop (SiL) | Enables dynamic representation of the transportation system, evaluation of mobility scenarios, and optimization of traffic conditions through real-time simulation. |
Frameworks based on the Internet of Things (IoT) represent the initial layer within smart ecosystems, as they enable the continuous acquisition of information through sensors, connected devices, and monitoring systems. In the context of urban transportation, these technologies facilitate the collection of variables related to traffic flow, infrastructure conditions, vehicle location, and travel patterns. However, the massive generation of data requires additional mechanisms capable of processing and transforming this information into useful knowledge for decision-making.
Artificial intelligence-oriented frameworks, such as the CityAI model, represent an evolution toward systems capable of interpreting large volumes of information and generating adaptive responses to changes in the urban environment. These models incorporate machine learning and computational analysis techniques to identify patterns, optimize processes, and improve the management of Intelligent Transportation Systems. Their main strength lies in their ability to transform historical data and real-time data into useful information for transportation planning and operations.
Distributed architectures based on edge computing, fog computing, and federated models complement these approaches by enabling information processing to take place closer to the data source. This is essential in smart mobility applications, where rapid response times and reduced latency are critical factors for analyzing dynamic traffic conditions. Additionally, these models help improve aspects related to security, privacy, and computational efficiency.
Studies related to digital twins (Digital Twin) show a trend toward the creation of dynamic virtual representations of urban systems. Frameworks such as Snap4City and digital twin-based models using microsimulation demonstrate that these technologies enable the integration of heterogeneous information, the representation of real-world environmental conditions, and the evaluation of scenarios before implementing changes to physical systems. In the field of transportation, these capabilities make it possible to analyze optimization strategies, simulate vehicle behavior, and improve sustainable mobility planning.
The comparison conducted reveals that technological evolution within smart cities does not occur through the replacement of one approach with another, but rather through the progressive integration of different paradigms. Each framework contributes a specific capability within the urban ecosystem: IoT enables the capture of information; artificial intelligence enables the analysis of patterns and the generation of predictions; distributed architectures provide efficient processing; while digital twins and governance models enable the strategic representation, management, and use of information.
This integration is particularly relevant for systems designed to identify areas of traffic congestion, as these systems require combining different sources of information, such as GPS trajectories, urban sensors, historical records, and real-time data. The combined application of artificial intelligence techniques, spatiotemporal analysis, and digital architectures enables the development of models capable of identifying mobility patterns, detecting recurring behaviors, and supporting traffic management decisions.
Future intelligent transportation systems are moving toward hybrid architectures where different technologies work in a complementary manner. The combination of data capture via the IoT, processing using artificial intelligence, trajectory analysis, and appropriate information management mechanisms represents a highly promising approach for improving our understanding of urban mobility and developing more efficient, sustainable, and adaptive solutions to today’s transportation challenges.
5. Discussion
The results obtained in this study show that the development of smart urban systems has evolved significantly in recent years, establishing itself as a multidisciplinary field driven by the convergence of emerging technologies. The sustained growth in scientific output identified in the bibliometric analysis confirms that smart cities have become a central focus of research, in line with recent studies that highlight their relevance to the digital transformation of urban environments [3,11,44].
One of the most significant findings concerns the predominance of hybrid architectures that integrate multiple technologies, particularly the Internet of Things, artificial intelligence, and distributed computing. This result aligns with what has been reported in the literature, which indicates that current urban solutions require integrated approaches to address the complexity of modern urban systems [7,8,48]. In this regard, the classification proposed in this study reinforces the idea that frameworks should not be analyzed in isolation, but rather as interdependent components within broader technological ecosystems.
With regard to artificial intelligence techniques, there is a shift from traditional machine learning models toward more advanced approaches based on deep learning and reinforcement learning. This evolution responds to the need to process large volumes of urban data and generate adaptive responses in real time. However, despite their advantages, these models have significant limitations in terms of interpretability and resource consumption, which aligns with concerns highlighted in recent research [23,45,46].
The analysis also reveals that, although IoT-based architectures remain fundamental, their effectiveness depends largely on the quality of the underlying infrastructure. Issues such as interoperability between devices, technological heterogeneity, and the management of large volumes of data continue to pose critical challenges. These findings are consistent with studies highlighting the need to establish common standards and improve integration among urban platforms [5,47].
The adoption of distributed architectures such as edge and fog computing reflects a direct response to the limitations of centralized cloud models, particularly in applications that require low latency and real-time processing. However, their implementation introduces new complexities related to the management of distributed nodes and data synchronization, suggesting the need for further research into efficient coordination and control mechanisms in decentralized environments [33,35].
From a broader perspective, the results also highlight a growing concern for non-technological aspects, such as sustainability, governance, and ethics in data use. This approach demonstrates a maturation of the smart city concept, shifting from a technology-centric vision toward a more comprehensive model focused on citizen well-being. In this context, recent research underscores the importance of incorporating principles of transparency, privacy, and citizen participation into the design of smart urban systems [41,51].
Despite the advances identified, significant limitations persist both in the literature and in this study. First, most of the studies analyzed focus on urban environments with a high level of technological development, which limits the generalizability of the results to contexts with limited resources. Second, the diversity of approaches and the lack of standardization make it difficult to directly compare proposals, highlighting the need for common evaluation frameworks.
The findings of this research suggest that the future of smart urban systems will be shaped by the integration of technologies, automation based on artificial intelligence, and the adoption of sustainable approaches. However, to achieve this goal, it will be necessary to overcome technical, organizational, and ethical barriers, as well as to promote closer collaboration among academia, industry, and the public sector.
6. Conclusions
This research enabled a systematic analysis of the state of the art in frameworks for adaptive smart urban systems, based on a structured review of the scientific literature. By applying the PRISMA model, we were able to refine an initial set of 2,528 records to a final sample of 64 relevant articles, thereby ensuring the quality, relevance, and methodological rigor of the study.
The results show that research on smart cities has experienced sustained growth in recent years, driven primarily by advances in technologies such as artificial intelligence, the Internet of Things, and distributed computing. This increase reflects global interest in developing innovative solutions to optimize urban management and improve quality of life in increasingly complex environments.
One of the main contributions of this study lies in the classification of the frameworks identified in the literature, which were grouped into four categories: IoT-based architectures, AI-driven models, distributed architectures, and approaches focused on sustainability and governance. This classification revealed that there is no single dominant approach, but rather a trend toward the integration of multiple technologies into hybrid architectures.
Furthermore, it was found that smart urban systems have evolved from models focused on data collection toward more advanced approaches that incorporate predictive analytics, automation, and adaptive capabilities. In this context, artificial intelligence plays a fundamental role by enabling data-driven decision-making and the dynamic optimization of urban services.
However, the study also highlights the existence of significant challenges that limit the effective implementation of these systems. Among these are the lack of interoperability between platforms, scalability issues, reliance on high-quality data, as well as concerns related to privacy, security, and information governance.
Based on these findings, it is concluded that the development of adaptive smart urban systems requires not only technological advances but also comprehensive approaches that take social, economic, and regulatory aspects into account. The adoption of common standards, the strengthening of digital infrastructure, and the incorporation of ethical principles will be key factors in ensuring the success of future implementations.
Finally, as a future line of work, we propose the development of integrated frameworks that combine artificial intelligence, distributed computing, and sustainability, as well as the validation of these models in real urban contexts. We also recommend further study of emerging techniques such as digital twins, explainable artificial intelligence, and autonomous systems, with the aim of moving toward truly smart, resilient, and citizen-centered cities.
Author Contributions
Conceptualization, G.R. and R.T.-B.; methodology, G.R., J.R. and D.R.; software, G.R. and J.R.; validation, R.T.-B., L.L. and W.H.; formal analysis, G.R., D.R. and J.B.-M.; investigation, G.R., J.R. and D.R.; resources, R.T.-B., C.G.-R., L.L. and W.H.; data curation, G.R. and J.R.; writing—original draft preparation, G.R.; writing—review and editing, R.T.-B., L.L., W.H. and C.G.-R.; visualization, G.R. and J.R.; supervision, R.T.-B., C.G.-R., L.L. and W.H.; project administration, G.R. and D.R.; funding acquisition, R.T.-B. All authors have read and agreed to the published version of the manuscript.
Funding
The authors would like to thank the Office of the Vice Rector for Research and Social Innovation at the Bolivarian University of Ecuador for its financial support, which contributed to the development and publication of this work.
Institutional Review Board Statement
Not applicable.
Informed Consent Statement
Not applicable.
Data Availability Statement
The dataset (specify type) supporting this (article/paper) has been deposited on Zenodo and is publicly available under the Creative Commons Attribution 4.0 International license, CC BY 4.0. The dataset can be accessed via the following DOI: https://doi.org/10.5281/zenodo.20801866 (accessed on June 22, 2026)
Acknowledgments
The authors would like to thank the Latin American Institute for the Future of Education for the methodological support provided through the WISE: AI Literacy Hub (PROY-INB-UBE-030) research network at the Bolivarian University of Ecuador, whose guidance contributed to the development of this study. We also thank the Artificial Intelligence Research Group at the Bolivarian University of Ecuador (GIIA) for its technical support.
Use of Artificial Intelligence
During the preparation of this manuscript, the authors used ChatGPT (GPT-5.3, OpenAI) for language editing, text organization, and to improve clarity and readability. The authors have reviewed and edited the output and take full responsibility for the content of this publication.
Conflicts of Interest
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
References
- Adreani, L.; Bellini, P.; Fanfani, M.; Nesi, P.; Pantaleo, G. Smart City Digital Twin Framework for Real-Time Multi-Data Integration and Wide Public Distribution. IEEE Access 2024. [Google Scholar] [CrossRef]
- Channi, H.K.; Kumar, P.; Safdarian, F. Planning and building digital twins for smart cities. In Digital Twins for Smart Cities and Villages; 2024. [Google Scholar] [CrossRef]
- Bittencourt, J.C.N.; Costa, D.G.; Portugal, P.; Vasques, F. A Survey on Adaptive Smart Urban Systems. IEEE Access 2024. [Google Scholar] [CrossRef]
- Gupta, S.; Kohli, R. AIoT for real-time traffic analysis and optimisation. AIoT Smart Transp. Transform. Future Mobil. 2025. [Google Scholar] [CrossRef]
- Bozkurt, Y.; Rossmann, A.; Pervez, Z.; Ramzan, N. Assessing data governance models for smart cities: Benchmarking data governance models on the basis of European urban requirements. Sustain. Cities Soc. 2025. [Google Scholar] [CrossRef]
- Hafez, W. Approach for defining intelligent systems technical performance metrics. Performance Metrics for Intelligent Systems (PerMIS) Workshop, 2012. [Google Scholar] [CrossRef]
- Alourani, A.; Alam, M.; Ali, A.; Khan, I.R.; Samal, C.K. Hybrid AI-IoT Framework with Digital Twin Integration for Predictive Urban Infrastructure Management in Smart Cities. Comput. Mater. Contin. 2026. [Google Scholar] [CrossRef]
- Rathod, D.B.; Prakash, D.; Harshavardhan, K.P.; Mahajan, S.; Jayanthi, S.; Fiza, S. AI-Driven Decision-Making Architectures for Future-Ready Smart Urban Infrastructures. In 2025 IEEE 5th International Conference on ICT in Business Industry and Government, ICTBIG 2025; 2025. [Google Scholar] [CrossRef]
- Banciu, C.; Florea, A. AIoT at the Frontline of Climate Change Management: Enabling Resilient, Adaptive, and Sustainable Smart Cities. Climate 2026. [Google Scholar] [CrossRef]
- Flores-Robles, A.E.; Silva-Carrillo, A.G.; Maliza-Muñoz, W.F.; Reyes-Zambrano, G.X. Educaplay para la mejora de la comprensión lectora en estudiantes de quinto grado de primaria. Rev. Científica Zambos 2025, 4, 21–37. Available online: https://scholar.google.com. [CrossRef]
- Hota, L.; Nayak, B.P.; Jain, P.K.; Kumar, A. Recent Trends in Artificial Intelligence Applications for Smart Cities: A Review. In Emerging Electrical and Computer Technologies for Smart Cities: Modelling, Solution Techniques and Applications; 2024. [Google Scholar] [CrossRef]
- Praharaj, S. Combining AI with urban resilience planning: A bibliometric analysis of literature. City Environ. Interact. 2026. [Google Scholar] [CrossRef]
- Jeyalakshmi, J.; Vijay, K.; Berna, E.; Manohar, E. Smart cities and urban sustainability with AI. Converg. Self-Sustain. Syst. With AI IoT 2024. [Google Scholar] [CrossRef]
- Kuberkar, S.; Singh, S.; Singhal, T.K. Adoption intention of artificial intelligence enabled smart city services from citizens’ perspective. Int. J. Hum. Cap. Urban Manag. 2024. [Google Scholar] [CrossRef]
- Durga, K.S.; Murty, A.V.N.; Kesalu, S.V.; Chandra, D.; Mishra, P.; Upadhyay, M.; Rao, A. Smart Cities And Social Equity: A Data Science Perspective On Urban Development. J. Appl. Bioanal. 2025. [Google Scholar] [CrossRef]
- Saritha, G.; Hassan, M.M.; Tamrakar, G.; Ganapathi Raju, N.V.; Velmurugan, R.; Al-Dulaimi, H.W.; Algburi, S. SmartCityNext: Leveraging AI and Machine Learning for Responsive and Efficient Urban Services. ICCR 2025 - 3rd International Conference on Cyber Resilience; 2025. [Google Scholar] [CrossRef]
- Lanzarini, L.C.; Hasperué, W.; Villa Monte, A. Minería de datos, minería de textos y Big Data. In Proceedings of the XXI Workshop de Investigadores en Ciencias de la Computación, 2019; Available online: https://scholar.google.com.
- Reyes, G.; Lanzarini, L.; Hasperué, W.; Bariviera, A.F. GPS Trajectory Clustering Method for Decision Making on Intelligent Transportation Systems. J. Intell. Fuzzy Syst. 2020, 38, 5529–5535. Available online: https://scholar.google.com. [CrossRef]
- Aydogdu, B.; Salah, A.A. Machine learning for urban computing. In Machine Learning and the City: Applications in Architecture and Urban Design; 2022. [Google Scholar] [CrossRef]
- Huszak, A.; Simon, V.; Bokor, L.; Tizedes, L.; Pekar, A. An AI-Driven Intelligent Transportation System: Functional Architecture and Implementation. Infocommunications J. 2024. [Google Scholar] [CrossRef]
- Koduri, V.; Srinivasan, D.; Parameswari, S.; Kiran, A.; Majeed, J.H.; Mahdi, S.A.; Alhayaly, B.H. SmartCityPredict: An AI and Machine Learning Framework for Enhancing Urban Service Efficiency Through Real-Time Data Analytics. ICCR 2025 - 3rd International Conference on Cyber Resilience; 2025. [Google Scholar] [CrossRef]
- Becerik-Altindis, S.; Ayatac, H. Artificial Intelligence in Urban Planning: Co-Word Analysis for Emerging Topics. J. Des. Built Environ. 2025. [Google Scholar] [CrossRef]
- Si, S.; Yao, Y.; Wu, J. Artificial Intelligence in Urban Planning: A Bibliometric Analysis and Hotspot Prediction. Land 2025. [Google Scholar] [CrossRef]
- Rejeb, A.; Rejeb, K.; Zaher, H.F.; Simske, S. Blockchain and Smart Cities: Co-Word Analysis and BERTopic Modeling. Smart Cities 2025. [Google Scholar] [CrossRef]
- Reyes, G.; Lanzarini, L.; Estrebou, C.; Bariviera, A.F. Dynamic Grouping of Vehicle Trajectories. J. Comput. Sci. Technol. 2022, 22, 141–150. Available online: https://scholar.google.com. [CrossRef]
- Reyes, G.; Lanzarini, L.; Hasperué, W.; Bariviera, A.F. Proposal for a Pivot-Based Vehicle Trajectory Clustering Method. Transp. Res. Rec. 2022, 2676, 281–295. Available online: https://scholar.google.com.
- Choudhary, S.; Kumar, K.; Gupta, A.K.; Sharma, H.; Hewage, P.; Harinda, E.; Iwendi, C. Advancements in Traffic Signal Control: A Review of Reinforcement Learning and Deep Learning Approaches. In Lecture Notes in Networks and Systems; 2025. [Google Scholar] [CrossRef]
- Thadikamalla, S.; Joshi, P.; Gangadharan, D. Reinforcement learning for traffic signal control: advancing efficiency through hybrid exploration strategies. J. Supercomput. 2025. [Google Scholar] [CrossRef]
- Dhatchayani, K.; Shubavathy, R.; Reshma, G.R.; Amalya Eunice, A.; Sundar, D.; Vezhaventhan, D. Optimizing Smart City Infrastructure Using 5G Edge AI with Adaptive Multi-Agent Reinforcement Learning. In Proceedings of International Conference on Visual Analytics and Data Visualization, ICVADV 2025; 2025. [Google Scholar] [CrossRef]
- Reyes, G.; Estrada, V.; Tolozano-Benites, R.; Maquilón, V. Batch Simplification Algorithm for Trajectories over Road Networks. ISPRS Int. J. Geo-Inf. 2023, 12, 399. [Google Scholar] [CrossRef]
- Mathew, P.L.; Chaudhry, S.; Upreti, K.; Farooqui, S.; Radhakrishnan, G.V. Implementing smart cyber-physical systems in industrial and urban applications: A practical approach. In Cognitive Cloud Computing: Building Intelligent Systems for Tomorrow; 2025. [Google Scholar] [CrossRef]
- Swetha, G.; Shruthi, M.; Veena, M.S.; Vishwanatha, S.; Krishnamurthy, T.; Raghu, N.; Kumar, S.D.; Trupti, V.N. The Convergence of Artificial Intelligence and the Internet of Things for Next-Generation Smart Systems. In Navigating Cybersecurity and Privacy in the Evolution of Smart Urban Ecosystems; 2026. [Google Scholar] [CrossRef]
- Verma, R.K.; Kishor, K.; Galletta, A. Federated Learning Shaping the Future of Smart City Infrastructure. In Federated Learning for Smart Communication using IoT Application; 2024. [Google Scholar] [CrossRef]
- Reyes, G.; Tolozano-Benites, R.; Lanzarini, L.; Estrebou, C.; Bariviera, A.F.; Barzola-Monteses, J. Methodology for the Identification of Vehicle Congestion Based on Dynamic Clustering. Sustainability 2023, 15, 16575. [Google Scholar] [CrossRef]
- Uddin, M.R.; Shankar, G.; Islam, S. Federated learning-based edge intelligence for IoT. In Edge Intelligence and Analytics for Internet of Things; 2025. [Google Scholar] [CrossRef]
- Sheelam, G.K. Federated Edge Intelligence: Enabling Privacy-Preserving AI for Smart Cities and IoT Systems. MSW Manag. 2024. [Google Scholar]
- Islam, A.; Karimipour, H.; Gadekallu, T.R. TwinSnake: A ZTN-Orchestrated Architecture for Secure AIoT Model Training with Digital Twins and Bio- Inspired Snake Learning in Smart Cities. In Proceedings - 2025 IEEE International Conference on Cloud Computing Technology and Science, CloudCom 2025; 2025. [Google Scholar] [CrossRef]
- Tepjit, S. Conceptualization of an Integrated Multi-Agent Framework for Innovative Urban Planning. Procedia Comput. Sci. 2025. [Google Scholar] [CrossRef]
- Lis, M.; Madziel, M. Green Transportation Planning for Smart Cities: Digital Twins and Real-Time Traffic Optimization in Urban Mobility Networks. In Applied Sciences (Switzerland); 2026. [Google Scholar] [CrossRef]
- Alsuwaiket, M.A. Optimizing Autonomous Vehicle Navigation Through Reinforcement Learning in Dynamic Urban Environments. World Electr. Veh. J. 2025. [Google Scholar] [CrossRef]
- Mupfumira, P.; Mutingi, M.; Sony, M. Smart city frameworks SWOT analysis: a systematic literature review. Front. Sustain. Cities 2024. [Google Scholar] [CrossRef]
- Pechtor, V.; Basl, J. ANALYSIS OF SUITABLE FRAMEWORKS FOR ARTIFICIAL INTELLIGENCE ADOPTION IN THE PUBLIC SECTOR. In IDIMT 2022 - Digitalization of Society, Business and Management in a Pandemic: 30th Interdisciplinary Information Management Talks; 2022. [Google Scholar] [CrossRef]
- Satpathy, I.; Nayak, A.; Jain, V. The Green City: Sustainable and Smart Urban Living Through Artificial Intelligence. In Utilizing Technology to Manage Territories; 2025. [Google Scholar] [CrossRef]
- Soni, L.; Taneja, A. Harnessing AI for Sustainable Smart Cities: Impact, Innovations, and Use Cases. In Artificial Intelligence (AI) for IT Energy Efficiency and Green AI for Environment Sustainability; 2026. [Google Scholar] [CrossRef]
- Khader, S.; Abu Al-Haija, Q. Explainable AI for Future Smart Cities: Architectures, Applications, and Challenges. In Explainable Artificial Intelligence for Trustworthy Decisions in Smart Applications; 2025. [Google Scholar] [CrossRef]
- Tong, S.; Li, S. Explainable Artificial Intelligence for Urban Planning: Challenges, Solutions, and Future Trends from a New Perspective. Int. J. Adv. Comput. Sci. Appl. 2024. [Google Scholar] [CrossRef]
- Xia, L.; Semirumi, D.T.; Rezaei, R. A thorough examination of smart city applications: Exploring challenges and solutions throughout the life cycle with emphasis on safeguarding citizen privacy. Sustain. Cities Soc. 2023. [Google Scholar] [CrossRef]
- Bibri, S.E.; Huang, J. Generative AI of things for sustainable smart cities: Synergizing cognitive augmentation, resource efficiency, network traffic, cybersecurity, and anomaly detection for environmental performance. Sustain. Cities Soc. 2025. [Google Scholar] [CrossRef]
- Balaji, C.G.; Menaka, S.; Rajeswari, G.; Ayyadurai, M.; Ponnusamy, S. Blockchain integrated federated learning for privacy preserving artificial intelligence and sustainable smart city infrastructure. In Smart City Computational Paradigms: A Sustainable Approach; 2026. [Google Scholar] [CrossRef]
- Antuley, U.; Siddiqui, S.; Hameed, S.; Arif, W.; Shah, S.A. SORA-ATMAS: Adaptive trust management and multi-LLM aligned governance for future smart cities. Knowl.-Based Syst. 2026. [Google Scholar] [CrossRef]
- Mirindi, D.; Sinkhonde, D.; Mirindi, F. An Advance Review of Urban-AI and Ethical Considerations. In Urban-AI 2024 - Proceedings of the 2nd ACM SIGSPATIAL International Workshop on Advances in Urban-AI; 2024. [Google Scholar] [CrossRef]
Table 2.
PRISMA-Based Study Selection Process
| Phase | Description | Number |
|---|---|---|
| Identification | Records retrieved from databases | 2528 |
| Screening | Filtering by title, abstract, and keywords | 105 |
| Eligibility | Full-text evaluation | 64 |
| Inclusion | Final studies selected | 64 |
Table 3.
IoT- and AIoT-based frameworks for smart urban systems.
| Framework | Authors | Year | Country/Context | Main Contribution |
|---|---|---|---|---|
| Hybrid AI-IoT Framework with Digital Twin Integration for Predictive Urban Infrastructure Management in Smart Cities | Alourani, Alam, Ali, Khan, and Samal | 2026 | International | Integrates IoT, artificial intelligence, and digital twins to anticipate failures and optimize predictive maintenance of urban infrastructure [7]. |
| AIoT for Real-Time Traffic Analysis and Optimization | Gupta and Kohli | 2025 | International | Uses sensors and connected devices along with intelligent algorithms to analyze traffic in real time, optimize urban mobility, and reduce traffic congestion [4]. |
| TwinSnake: A ZTN-Orchestrated Architecture for Secure AIoT Model Training with Digital Twins and Bio-Inspired Snake Learning in Smart Cities | Islam, Karimipour, and Gadekallu | 2025 | China, IEEE | Combines digital twins, bio-inspired learning, and Zero Trust Networking principles to protect the data used in AIoT model training [37]. |
| Implementing Smart Cyber-Physical Systems in Industrial and Urban Applications: A Practical Approach | Mathew, Chaudhry, Upreti, Farooqui, and Radhakrishnan | 2025 | International | Integrates sensors, IoT devices, and intelligent processing mechanisms to connect the physical environment with computational systems in urban and industrial applications [31]. |
| The Convergence of Artificial Intelligence and the Internet of Things for Next-Generation Smart Systems | Swetha et al. | 2026 | International | Analyzes the convergence of AI and IoT to build autonomous, efficient urban systems capable of dynamically responding to changing conditions [32]. |
Table 4.
AI-driven frameworks for smart urban systems.
| Framework | Authors | Year | Country/Context | Main Contribution |
|---|---|---|---|---|
| SmartCityPredict: An AI and Machine Learning Framework for Enhancing Urban Service Efficiency Through Real-Time Data Analytics | Koduri et al. | 2025 | International | Proposes an architecture based on AI and machine learning to transform urban data into predictions and recommendations aimed at improving the efficiency of public services [21]. |
| SmartCityNext: Leveraging AI and Machine Learning for Responsive and Efficient Urban Services | Saritha et al. | 2025 | India | Presents a smart platform to dynamically analyze urban needs and adapt services to changing environmental conditions [16]. |
| An AI-Driven Intelligent Transportation System: Functional Architecture and Implementation | Huszak, Simon, Bokor, Tizedes, and Pekar | 2024 | Hungary | Describes a functional architecture implemented to optimize urban mobility, traffic management, and real-time decision-making [20]. |
| AI-Driven Decision-Making Architectures for Future-Ready Smart Urban Infrastructures | Rathod et al. | 2025 | India | Proposes an architecture to automate decision-making in smart urban infrastructures through the analysis of complex scenarios [8]. |
| Explainable AI for Future Smart Cities: Architectures, Applications, and Challenges | Khader and Abu Al-Haija | 2025 | Jordan | Introduces explainable artificial intelligence to increase transparency, trust, and understanding of algorithmic decisions in smart cities [45]. |
| Conceptualization of an Integrated Multi-Agent Framework for Innovative Urban Planning | Tepjit | 2025 | Thailand | Proposes a multi-agent framework in which intelligent agents interact to analyze urban scenarios and support city planning [38]. |
Table 5.
Distributed, edge, and federated frameworks for smart urban systems.
| Framework | Authors | Year | Country/Context | Main Contribution |
|---|---|---|---|---|
| Federated Edge Intelligence: Enabling Privacy-Preserving AI for Smart Cities and IoT Systems | Sheelam | 2024 | International | Combines edge computing and AI to process data close to the source, reduce centralization, and preserve the privacy of urban information [36]. |
| Federated Learning-Based Edge Intelligence for IoT | Uddin, Shankar, and Islam | 2025 | International | Integrates federated learning and edge intelligence to train distributed models without transferring raw data to central servers [35]. |
| Federated Learning Shaping the Future of Smart City Infrastructure | Verma, Kishor and Galletta | 2024 | International | Explores federated learning as a mechanism to improve scalability, collaboration among entities, and decentralized management of urban data [33]. |
| Optimizing Smart City Infrastructure Using 5G Edge AI with Adaptive Multi-Agent Reinforcement Learning | Dhatchayani et al. | 2025 | India/global | Integrates 5G, Edge AI, and multi-agent reinforcement learning to optimize urban infrastructure and respond in real time to dynamic events [29]. |
| Blockchain-Integrated Federated Learning for Privacy-Preserving Artificial Intelligence and Sustainable Smart City Infrastructure | Balaji et al. | 2026 | India/global | Combines blockchain and federated learning to strengthen privacy, traceability, and trust in distributed urban ecosystems [49]. |
Table 6.
Sustainability- and governance-oriented frameworks for smart urban systems.
| Framework | Authors | Year | Country/context | Main contribution |
|---|---|---|---|---|
| Smart City Digital Twin Framework for Real-Time Multi-Data Integration and Wide Public Distribution | Adreani, Bellini, Fanfani, Nesi, and Pantaleo | 2024 | Italy | Integrates multiple urban data sources in real time using digital twins and facilitates the public distribution of information for city management [1]. |
| Assessing Data Governance Models for Smart Cities | Bozkurt, Rossmann, Pervez, and Ramzan | 2025 | Europe | Compares data governance models and evaluates their ability to address regulatory, ethical, and operational requirements of European smart cities [5]. |
| SORA-ATMAS: Adaptive Trust Management and Multi-LLM Aligned Governance for Future Smart Cities | Antuley et al. | 2026 | International | Introduces adaptive trust management and governance aligned with multiple large language models (LLMs) to strengthen the reliability of smart urban systems [50]. |
| Green Transportation Planning for Smart Cities: Digital Twins and Real-Time Traffic Optimization in Urban Mobility Networks | Lis and Madziel | 2026 | Poland | Uses digital twins and real-time traffic optimization to improve urban mobility and reduce the environmental impacts of transportation [39]. |
| AIoT at the Frontline of Climate Change Management: Enabling Resilient, Adaptive, and Sustainable Smart Cities | Banciu and Florea | 2026 | International | Analyzes the use of AIoT to strengthen urban resilience, climate adaptation, and sustainability in the face of environmental phenomena [9]. |
| Harnessing AI for Sustainable Smart Cities: Impact, Innovations, and Use Cases | Soni and Taneja | 2026 | International | Examines AI applications aimed at energy efficiency, resource optimization, and reducing environmental impacts [44]. |
| The Green City: Sustainable and Smart Urban Living Through Artificial Intelligence | Satpathy, Nayak, and Jain | 2025 | International | Highlights the potential of AI to improve resource management, the efficiency of public services, and sustainable urban development [43]. |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2026 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
Copyright: This open access article is published under a Creative Commons CC BY 4.0 license, which permit the free download, distribution, and reuse, provided that the author and preprint are cited in any reuse.
