Advancing Smart City Sustainability Through Artificial Intelligence, Digital Twin and Blockchain Solutions

Abstract

This paper presents an integrated Smart City platform that combines digital twin technology, advanced machine learning, and a private blockchain network to enhance data-driven decision making and operational efficiency in both public enterprises and small and medium-sized enterprises (SMEs). The proposed cloud-based business intelligence model automates Extract, Transform, Load (ETL) processes, enables real-time analytics, and secures data integrity and transparency through blockchain-enabled audit trails. By implementing the proposed solution, Smart City and public service providers can significantly improve operational efficiency, including a 15% reduction in costs and a 12% decrease in fuel consumption for waste management, as well as increased citizen engagement and transparency in Smart City governance. The digital twin component facilitated scenario simulations and proactive resource management, while the participatory governance module empowered citizens through transparent, immutable records of proposals and voting. This study also discusses technical, organizational, and regulatory challenges, such as data integration, scalability, and privacy compliance. The results indicate that the proposed approach offers a scalable and sustainable model for Smart City transformation, fostering citizen trust, regulatory compliance, and measurable environmental and social benefits.

1. Introduction

In today’s digital economy, the need for effective business intelligence (BI) systems is growing exponentially [1], especially given the huge amount of data generated every day. However, while business intelligence offers significant benefits in terms of data-driven decision making, the implementation of these systems in small- and medium-sized enterprises (SMEs) faces numerous challenges such as limited scalability, high implementation costs, outdated data, security threats, lack of management support, insufficient financial resources, ineffective project management, and resistance to change, all of which can hinder their successful adoption and use of advanced business intelligence systems. These threats make it difficult for SMEs to access business intelligence, while the cost is often the biggest barrier.

Smart Cities can act as data hubs, offering ETL processes, analytics, and advisory services to their public enterprises and departments, thus supporting holistic digital transformation and operational excellence across the urban ecosystem [2]. In addition to challenges and opportunities for SMEs, this paper presents the integration of digital twin technology, artificial intelligence to secure scalable data services, not only to private SMEs, but also to public enterprises and city administrations.

According to the research [3], SMEs face a number of obstacles when adopting business intelligence systems. One of the key challenges is exceeding the implementation budget since they do not have sufficient financial resources to launch projects. Also, ineffective BI project management and a lack of management support are obstacles to the successful implementation of BI systems in SMEs. A lack of understanding of BI system capabilities among key stakeholders leads to uncertainty and resistance to change, globalization, and a turbulent environment, where finding competitive advantages is a difficult task for both small and large enterprises [4]. Similar problems are identified in [5], where the authors state that the lack of training and support for users is another significant challenge. SMEs often do not have the necessary resources to train their employees in the use of BI tools, which can lead to a low adoption rate of the system. In addition, user resistance to change is often present, as many employees believe that BI systems require significant changes in their work processes and decision-making methods. However, despite the challenges, the implementation of BI systems in SMEs brings numerous advantages. SMEs that successfully implement BI systems report greater decision-making efficiency. BI systems enable faster data analysis and better understanding of market trends, which enables SMEs to make better strategic decisions based on data.

Security and data protection are also key factors for SMEs when digitizing their business using data science and artificial intelligence, and it is necessary to use advanced security measures to protect privacy [6]. In order to increase security, transparency, and data protection, this paper suggests connecting the BI system with a private blockchain network. This approach allows SMEs to protect themselves from threats, thereby increasing trust between partners and clients. Although the implementation of a BI system in SMES can be challenging, considering financial, human, and technical resources, the benefits it provides in the form of better data analysis, increased efficiency, and data security make BI a key tool for digital transformation. By optimizing business processes, reducing costs, and improving decision making, SMEs can achieve a competitive advantage in the global market.

This paper presents a cloud-based business intelligence model that integrates advanced machine learning algorithms into the ETL process, proposes a digital twin of the Smart City, and utilizes a private blockchain network. The functionalities of the proposed Smart City digital twin include automating data collection and preparation, while advanced analytics and predictive models based on machine learning are applied to optimize urban processes. At the same time, private blockchain technology is employed to ensure confidentiality, integrity, and transparency of data to address key security challenges in business analytics. The digital twin facilitates real-time simulation and proactive management of urban resources, supporting scenario analysis and data-driven decision making for public enterprises and city administrations. Through blockchain-enabled audit trails, the integrity and traceability of all ETL process stages are maintained, supporting regulatory compliance. The functionalities of the proposed Smart City model are demonstrated in a pilot implementation, where measurable improvements in operational efficiency, cost reduction, environmental impact, and citizen engagement are achieved, confirming the potential of the integrated digital twin and blockchain approach for sustainable Smart City transformation.

While recent research has demonstrated the potential of digital twins, AI, and blockchain to enhance various aspects of Smart City management—including real-time simulation, participatory governance, and secure data sharing—most existing solutions focus on isolated applications or lack integration across the full urban data lifecycle [7,8]. The approach presented in this paper distinguishes itself by delivering a unified, modular platform that tightly couples a real-time digital twin with advanced machine learning–driven ETL processes and a permissioned blockchain infrastructure for end-to-end data integrity, transparency, and participatory governance. Unlike prior studies that emphasize either digital twin analytics, blockchain-secured audit trails, or AI-powered optimization in isolation, our model uniquely enables seamless interoperability between public services, SMEs, and citizens through open APIs and privacy-preserving identity management [8,9]. This comprehensive integration not only supports automated, secure, and transparent urban data workflows but also empowers stakeholders—including city administrations, enterprises, and citizens—to collaboratively co-create, monitor, and govern Smart City services in real time. The pilot deployment demonstrates measurable benefits in operational efficiency, citizen engagement, and environmental impact, positioning the proposed solution as a scalable and replicable model for next-generation sustainable smart cities.

Despite significant advances in Smart City technologies, several critical challenges remain unaddressed. First, data interoperability is a persistent issue, as urban data originates from diverse sources and different formats, which complicates integration and analysis. Second, security and privacy concerns arise due to the sensitive nature of urban and citizen data, necessitating robust mechanisms for data protection and compliance. Third, effective stakeholder engagement, including citizen participation and transparent governance, remains limited by the lack of trustworthy and accessible digital platforms. This paper addresses these challenges by proposing an integrated Smart City platform that combines digital twin technology, advanced machine learning, and private blockchain infrastructure to ensure seamless data interoperability, enhanced security, and participatory governance.

The paper is structured as follows: Section 2 reviews related work on the integration of digital twin technology, AI, and blockchain in Smart City. Section 3 introduces the proposed digital twin and blockchain architecture. Section 4 describes the implementation of the private blockchain network using the Hyperledger Besu. Section 5 presents the evaluation of the pilot implementation and discusses technical and organizational challenges as well as directions for future work. Section 6 concludes the paper.

2. Related Work

Recent research highlights the transformative potential of integrating real-time digital twins, blockchain-enabled participatory governance, and AI-powered resource management at the city scale [7,9]. By leveraging these technologies, cities are enabled in real time to simulate, monitor, and optimize core systems such as energy, water, waste management, and mobility. Digital twins serve as dynamic, data-driven replicas of city environments, allowing for scenario modeling, predictive analytics, and proactive resource allocation [10,11]. Blockchain technology is being deployed to underpin participatory governance, providing immutable and transparent records of citizen proposals, voting, and administrative actions [12,13]. This ensures that all stakeholders can audit decision-making processes and fosters greater trust in public administration. AI-driven analytics further enhance operational efficiency by enabling predictive maintenance, demand forecasting, and the automation of complex city processes, leading to measurable improvements in cost savings, environmental impact, and service quality. Open dashboards and transparent data platforms are used to engage citizens, offering real-time insights into city performance and enabling meaningful participation in city governance [10,14,15]. The combination of these technologies not only improves operational excellence but also supports regulatory compliance, sustainability goals, and citizen empowerment, positioning cities as resilient, data-driven ecosystems ready to address future challenges.

The ETL process is a key step in the data preparation for analytical systems, whereby data from various sources is collected, processed, and loaded into data warehouses. Traditional ETL systems often face challenges such as handling large amounts of data, identifying anomalies, ensuring data quality, and supporting real-time updates. In this context, machine learning brings significant innovations, enabling the automation and optimization of these processes in order to improve the performance and accuracy of data processing [16,17].

Machine learning and decision making require a large volume of heterogeneous data in various complex formats. To solve this problem, a hybrid swarm intelligence algorithm with tabu search for handling large amounts of data in a cloud-based ETL process in a data warehouse is proposed [18]. Furthermore, in [19], the authors provide an overview of the existing ETL process and data quality approaches to enable efficient data processing by sharing data in a distributed cloud architecture. The data is collected from various sources such as the private sector, companies, government sectors, and their analysis helps users to make better business and organizational decisions. The data collection can be optimized by determining the time intervals for data collection from external sources [20]. Another approach is proposed in [21], where the authors propose various ETL tools for data processing when making business decisions. Business intelligence is widely applied across many domains, including healthcare for patient data analysis and decision support data [22], as well as for real-time data processing and secure analytics using relational algebra and other advanced techniques [23].

Traditional ETL processes often face several key challenges, such as a lack of scalability, where large amounts of data can slow down processing, which is especially problematic in big data environments. Traditional ETL systems are often not scalable enough to handle large volumes of data [24,25]. Also, handling and integration of heterogeneous data sources, which can be unstructured and semi-structured data, traditional systems cannot always process efficiently [26]. Manual anomaly detection and correction of errors in data is a time-consuming process, which can negatively affect data quality and analysis accuracy [27]. Real-time updates pose a challenge for many traditional ETL systems, which are optimized for periodic data loads [28]. The integration of machine learning into ETL processes enables significant improvements in the following stages:

• Data extraction: Machine learning algorithms can automatically identify key information from large and diverse data sources, including text documents, Internet of Things (IoT) devices, and social networks. Automating this step reduces the time required for data collection and increases process reliability [16,29].
• Data transformation: Algorithms such as clustering and deviation analysis can recognize irregularities in data and automatically correct them, thereby improving data quality and standardization of data from heterogeneous sources [30]. Prediction of missing values by techniques such as regression allows filling in missing data based on existing patterns, reducing the need for manual data entry or discarding.
• Data loading: By optimizing performance, machine learning algorithms can predict optimal time intervals for loading large amounts of data, ensuring minimal system load and faster processing [6,31]. Continuous updating and synchronization of data is enabled, which is crucial for applications that require updated information in real time [16].

Figure 1 illustrates the ETL process and the overall BI architecture, highlighting how data is collected from multiple sources, processed, and loaded into a data warehouse to support analytics and decision making in the Smart City platform.

By introducing advanced machine learning algorithms, ETL processes become faster, more reliable, and more adaptable to dynamic business needs. Machine learning algorithms for optimizing ETL processes represent a key step towards creating business intelligence systems that enable informed decisions based on high-quality and up-to-date data. This technology, with integration with blockchain for additional security, has the potential to revolutionize the way organizations process and use data, enabling them to successfully face the challenges of the digital age.

Blockchain technology, best known for its application in cryptocurrencies, is becoming increasingly relevant in other industry sectors, including business intelligence. Key features of blockchain, such as decentralization, non-repudiation, and transparency, make it ideal for solving security challenges in data analysis and processing systems. These advantages become even more pronounced in private networks because they allow better control over data within a closed ecosystem. The application of blockchain technology ensures the permanent recording of transactions or changes in data sets. Once data is entered into the blockchain, it cannot be changed without leaving a trace. This feature ensures that the history of data can be easily verified, thus eliminating the possibility of manipulation [32]. For business intelligence, this means that data-driven analysis and decisions are always based on accurate and credible information [33]. Blockchain uses advanced encryption mechanisms to ensure the confidentiality and integrity of data. Only authorized users can access certain parts of the data within the network [34]. Private blockchain networks further restrict access to only authorized participants, reducing the risk of unauthorized intrusion. In [35], the authors discuss and benchmark different blockchain platforms for data processing, and blockchain access control is explained to improve the security of private blockchain networks [36]. Furthermore, all transactions within the blockchain are visible to relevant participants, which allows for complete transparency and easier tracking of activities [37]. For example, in business intelligence systems, this can be used for audit trails to ensure that data is used and processed in accordance with regulations [12,38,39]. Unlike centralized systems, blockchain stores data on a distributed network, eliminating a single point of vulnerability [32]. Even if one node within the network becomes compromised, the rest of the system remains intact and functional [34].

For all of the above, the advantages of using blockchain in business intelligence are obvious because it can ensure data non-repudiation during all stages of the ETL process. During extraction, the blockchain can record metadata about the data source to ensure its authenticity, and during transformation, every change in the data is stored, creating an audit trail. When loaded, the data can be stored in a warehouse that is connected to the blockchain for continuous validation. Data protection regulations are also supported, as private blockchain networks can help organizations comply with regulations such as GDPR and HIPAA by ensuring transparency in how data is managed and providing evidence of compliance through a transaction log. Fraud and data misuse in business intelligence systems are prevented because data stored in the blockchain network can automatically detect attempts at manipulation or unauthorized access. Transparent logs within the blockchain can serve as evidence in the event of any disputes. Due to the application of smart contracts [40], blockchain can be applied for authentication and trust management in business intelligence [41], solving security problems and challenges [42,43], collecting data from IoT systems [44,45].

3. Integrated Digital Twin and Blockchain Architecture for Smart City Transformation

The integration of private blockchain technology with business intelligence brings a new level of security and transparency in data transformation and analysis. Data non-repudiation, encryption, transparency, and resistance to attacks enable Smart City organizations’ decision making with greater confidence. In combination with advanced machine learning algorithms and AI, blockchain becomes a key tool in the development of systems that provide reliable and secure business analytics, ready for the challenges of the digital age.

3.1. Business Intelligence Model and Digital Twin

The business intelligence model proposed in this paper has several key advantages compared to existing solutions:

• Increasing competitiveness: Using advanced machine learning algorithms in the ETL process enables more accurate data analysis. Users are provided with an insight into market trends, optimized business processes, and a better strategy, which achieves a competitive advantage on a global level [46].
• Digital business transformation: Integrating business intelligence in the cloud allows SMEs to modernize and migrate business processes to digital platforms by connecting them with blockchain technology to ensure data transparency and security, which is key to digital transformation in industries such as finance [47,48], healthcare [49], or retail [50].
• Increased security and trust: Blockchain networks offer a high level of security and transparency, which allows for easier verification of data and transactions. This increases trust among partners, customers, and regulators, which is key to the global competitiveness of SMEs.
• Scalability and flexibility: A cloud solution enables scalability, which means it can grow with the business while reducing the need for large initial infrastructure investments. It also uses the flexibility that clouds offer in terms of speed of implementation and adaptability to market changes [51,52].
• Optimizing business decisions: The use of business intelligence enables faster making of informed decisions based on data. With the help of advanced analytics and machine-based recommendations, Smart City companies can better understand market opportunities, reduce risks, and optimize resources, thus increasing their efficiency and reducing costs [53].

These advantages can significantly improve business processes and make SMEs more competitive on the global market, which is crucial for long-term sustainability and success in the digital era [54]. The proposed model extends the traditional BI approach by embedding a real-time digital twin of the Smart City by integrating IoT data streams (e.g., energy, water, waste, air quality, and mobility) and advanced visualization tools. This enables both Smart City officials and citizens to monitor urban metrics, participate in scenario planning, and create solutions for sustainable Smart City development. A digital twin enables the creation of a dynamic, real-time virtual replica of a city environment, process, or entire city and enables simulation of various scenarios to predict future outcomes. This virtual model is continuously synchronized with its real-world counterpart using data from sensors, IoT devices, and other sources, allowing for up-to-date monitoring, analysis, and simulation. By leveraging digital twins, organizations and city administration enables the following:

• Monitor and analyze the performance of assets and systems in real time, detecting inefficiencies or faults as they arise.
• Run simulations and scenario analyses to predict the outcomes of changes, optimize operations, and support data-driven decision making.
• Test new designs, policies, or interventions virtually before implementing them in the physical environment, reducing risk and costs.
• Enable predictive maintenance and proactive resource management, improving operational efficiency and reducing downtime.
• Foster transparency and participatory governance by providing stakeholders—including citizens—with access to real-time data and interactive dashboards.
• Support sustainability goals by optimizing resource usage, reducing emissions, and enhancing overall system performance.

In the context of Smart Cities, digital twins serve as a strategy accelerator, integrating static, historical, and real-time data to provide actionable insights into city administration, improve service delivery, and empower both decision makers and citizens.

3.2. Private Blockchain Network

The implementation of a private blockchain network requires detailed planning and implementation of specific steps to ensure security, reliability, and transparency. Private permissioned blockchain networks differ from public blockchain networks in that only authorized members have access to the network, which makes them an ideal solution for business purposes that require a high level of confidentiality and data control [35,36]. In the context of Smart Cities, blockchain is not only used for securing business intelligence data, but also for enabling participatory governance. In this paper, a permissioned blockchain infrastructure where all Smart City initiatives are immutably recorded, ensuring transparency, auditability, and trust, is proposed. Privacy-preserving digital identity management should be integrated to protect personal data and support secure authentication for both citizens and the Smart City administration.

Key steps in implementing a private blockchain network are choosing the right blockchain platform suitable for smart contracts [55]. The most commonly used platforms for this purpose include Hyperledger Fabric, which offers a modular architecture, flexibility, and support for smart contracts, making it suitable for complex business processes [56]. Corda focuses on data privacy and is used mainly in the financial sector [57]. Quorum is a version of the Ethereum blockchain network adapted for business use with an emphasis on privacy and transaction speed. The choice of platform depends on specific requirements such as performance, scalability, and support for smart contracts [58]. Hyperledger Besu is an EVM-compatible blockchain client that enables the development of private and consortium blockchain networks with a focus on security and [59].

In the next step, it is necessary to implement a Public Key Infrastructure (PKI) where certificates ensure that only authorized users can access the network. Authentication and authorization rules, where users and network nodes are given specific roles and access rights, thus minimizing the risks of abuse [34]. For example, network administrators have full access, while other users have limited privileges, depending on their role. Configuration of network nodes with different roles and functions, where validation nodes verify and approve transactions, and client nodes that enable users to access the network. Each node must be configured to communicate with other nodes via secure protocols. To achieve optimal performance, the network should be designed to ensure load balance between nodes [60]. Securing the network infrastructure is a key aspect of the security of a private blockchain network, including data encryption using asymmetric encryption to protect sensitive data [27]. Monitoring network activity using software tools to identify suspicious activities and unauthorized access attempts. Regular backups are required to ensure data integrity and enable rapid recovery after an incident.

After the network is secure, smart contracts should be implemented to enable the automation of business processes within the blockchain. For example, the authenticity of collected data can be verified where the data is checked and without the need for an intermediary. The blockchain stores metadata about all collected data in the ETL process, user queries, etc. Payment automation can be implemented after all conditions are met and the payment is executed automatically. Coding smart contracts requires thorough testing to prevent errors that could compromise the security or functionality of the network. Testing and optimization are key to uncovering weaknesses in the system. Performance testing checks the speed of transactions and the scalability of the network under different loads. It is necessary to provide security protocols that simulate potential attacks to ensure network security [61,62]. Compatibility with other systems participating in the model should also be checked. After testing and spotting potential flaws, the network is adjusted to achieve optimal performance. Maintenance and upgrades ensure security patches and regular software updates to eliminate security vulnerabilities. Continuous network monitoring to ensure reliability. Adding new features according to business needs. Maintenance requires an expert team to ensure that the network remains secure, scalable, and compliant with regulatory requirements [63].

3.3. Specific Model Requirements for Business Intelligence Applications

Blockchain networks are particularly useful in the context of business intelligence, where it is crucial to ensure data integrity throughout the ETL process. During the ETL process, the collected data is stored in a data warehouse [64]. A key innovation of the proposed digital twin model is the application of blockchain-based audit trails beyond the private sector, supporting compliance and transparency in public resource management (e.g., waste, mobility, energy). For example, smart waste bins equipped with IoT sensors, machine learning for route optimization, and blockchain for lifecycle tracking enable cities to reduce costs, improve recycling rates, and report transparently to citizens and regulators. The data most commonly stored in a data warehouse is as follows:

• Structured and unstructured data that are collected from various sources during the ETL process.
• Transformed data that is aggregated, normalized, and cleaned data ready for analytics.
• Metadata with information about the source, quality, and status of the data.

  ○

  Data quality metrics are often stored to monitor the accuracy, consistency, completeness, and timeliness of data. This data can include information about missing values, data irregularities, or input errors.

  ○

  Indexing is important for optimizing queries in data warehouses, enabling faster access to large amounts of data. Indexes help speed up data analysis, especially when the data is very voluminous.

  ○

  Data versioning can be stored to track the evolution and changes in data over time, providing a history of all data versions for later analysis.

  ○

  The transaction data can also be stored during the ETL process to track when and how data was extracted, transformed, and loaded. This data allows for audits and the detection of potential errors or problems in the ETL process.

  ○

  Reference data can be used to standardize data coming from different sources, such as country lists, industries, currencies, or product codes. This data ensures data consistency and accuracy.

  ○

  Business rules that define data validation and transformation can be stored to ensure that all data in the warehouse meets a defined set of conditions and requirements.

• Historical data needed for analysis and prediction models.

The metadata are also stored in the blockchain during all phases:

• During the data extraction phase, the following metadata is stored in the blockchain:

  ○

  Record ID, which is a unique identifier for each data item collected.

  ○

  Information about the source of the collected data (e.g., database, IoT device, API).

  ○

  Timestamps, the exact time of extraction to ensure historical traceability.

  ○

  Content of the collected data (e.g., measurements, transactions, and user data).

  ○

  Data hash values are calculated over key data sets to ensure their integrity.

  ○

  Information about the users or applications that initiated the extraction process.

  ○

  Records of the quality and validation of the data source (e.g., size, format, or the presence of anomalies).

• During the data transformation phase, the following metadata is stored in the blockchain:

  ○

  Transformation operations—Types of changes, such as aggregation, normalization, or data cleansing.

  ○

  Data versions—The history of changes for each transformation, including the original and transformed form of the data.

  ○

  Validation rules—The criteria applied to ensure the accuracy and consistency of the data.

  ○

  Error identification—Any anomalies or errors detected in the data, and how they were handled.

• During the data loading phase, the following metadata is stored in the blockchain:

  ○

  Destinations—Locations where the data is stored (e.g., a data warehouse or analytics platform).

  ○

  Load timestamps—Precise storage of the time of each load.

  ○

  Data integrity—Controls to ensure that data has not been modified during the loading process.

  ○

  Audit trail—A transparent record of all process steps, including any adjustments or interventions.

This implementation allows for the creation of a transparent and secure ETL process, thus ensuring confidence in analytical results and compliance with regulatory requirements [12]. It enables secure, transparent, and efficient processing while ensuring data confidentiality and integrity. Implementing these solutions facilitates the tracking of all transactions, analytical procedures, and historical records, which are essential for making informed decisions in cloud business intelligence. Figure 2 illustrates how data is extracted from diverse Smart City sources, processed, and transformed with the support of advanced machine learning and LLM algorithms, and stored in a warehouse. In each stage, blockchain technology is used to record metadata, ensure data integrity, and provide an immutable audit trail to guarantee transparency and trust in all business intelligence operations within the Smart City platform.

3.4. User Interaction with the Proposed Model Using Natural Language Processing

Once the data is stored in the data warehouse, users can ask the LLM Server queries related to the data in the warehouse, allow it to perform analysis from their own data, request advice, simulate possible scenarios, and more. In addition to SME users, the Smart City platform supports city administration, public enterprise managers, and citizens, enabling them to interact with the system via Natural Language Processing (NLP) queries, access real-time analytics, and participate in governance processes. All interactions are securely logged on the blockchain, ensuring auditability for both operational and participatory functions as illustrated in Figure 3. This process includes additional security and functional measures:

• User data should not be stored on the platform’s central servers to preserve the privacy of user data. Instead, the data remains on the user’s computer, and the LLM Server can only access it temporarily during processing.
• By recording all user queries and the results obtained by the LLM Server in the blockchain, authentication and transparency are enabled.
• It is possible to subsequently verify the results of the query against the system: the hash of the results allows for a later verification of the accuracy of the answers given by the system to the user. This can always be checked whether the system has advised the user correctly in the event of subsequent disputes or ambiguities. A security check can be made because the blockchain allows for the tracking of each step of the process, which ensures non-repudiation because it is impossible to retroactively change the data or results without leaving a trace. Transparency of the entire process is obtained because all relevant data about the process is available for audit.
• Audit trail: Implementing an audit trail allows for the tracking of user activities, when the data was used and for what purpose, and ensures that it is authentic.

What is stored in blockchain? Blockchain stores the following:

• The identifier of the user who made the query, which should be an anonymous or pseudonymous user ID.
• The timestamp of the user sending the query to the system.
• The query type, which indicates the category or type of query (e.g., prediction, trend analysis, and recommendations).
• Hash values of the data sent by the user during the data upload to the system to later used to verify the integrity of the data and ensure that it has not been tampered with.
• The query result is stored as a summary of the results, including the hash of the generated data or response.
• Audit record as a historical trace that includes all processing steps related to the query.

To clarify the AI layer of the system, the platform integrates several machine learning algorithms tailored to specific Smart City tasks. For predicting energy consumption, traffic management, waste generation, and optimizing collection schedules, Long Short-Term Memory (LSTM) neural networks can be used due to their proven ability to model temporal dependencies and non-linear trends in urban data streams. LSTM models are especially effective for capturing daily and weekly fluctuations in waste patterns, as confirmed in recent Smart City studies [65]. Random Forest classifiers can be applied to identify anomalies in sensor data (e.g., unexpected fill levels or sensor faults) and to classify waste types. Random Forests are mostly chosen for their robustness to noisy data and high performance in multi-class classification tasks, which is crucial for heterogeneous urban sensor environments [66]. It can also be used for route optimization and dynamic resource allocation are addressed using a hybrid approach that combines classical optimization algorithms with reinforcement learning. Reinforcement learning enables adaptive, real-time decision making in response to changing urban conditions. These algorithms are proposed based on a review of current best practices and testing with real-world data.

The AI layer of the system is enhanced by the integration of a dedicated LLM server, which enables advanced NLP capabilities for user interaction, analytics, and decision support. Once data is stored in the data warehouse, users, including SME managers, city administration, public enterprise managers, and citizens, can interact with the system via NLP-based queries. The LLM server should process these queries, perform analyses, provide recommendations, and simulate scenarios based on real-time and historical data. All user interactions are securely logged on the blockchain, ensuring auditability and transparency for both operational and participatory functions. The LLM server is designed to protect user privacy by not storing user data centrally. Instead, data remains on the user’s device, and the LLM server only accesses it temporarily during processing. Each query and its result are hashed and recorded as part of the blockchain’s immutable audit trail. This approach ensures that results can be verified at any time, and the integrity of the entire process is maintained. In addition to the LLM server, the system supports Large Language Model Agent integration (LAMA), which leverages the latest advancements in generative AI for Smart City management. The LAMA agent acts as an intelligent interface, enabling dynamic and context-aware communication between users and the system. It facilitates tasks such as automated report generation, anomaly detection, predictive analytics, and scenario simulation. The LAMA agent is deployed on the LLM server, which is provisioned with robust hardware and containerized using Docker for scalability and reliability. The server is managed through standard DevOps practices, including continuous integration, monitoring, and security updates. Integration with the broader Smart City platform could be achieved via open APIs, allowing seamless interoperability between the LLM server, digital twin, blockchain, and other city services. This modular architecture ensures that the LLM server and LAMA agent can be easily updated or replaced as new AI technologies emerge, supporting the long-term sustainability and adaptability of the system.

3.5. Comparative Analysis and Contextualization

Digital twins serve as dynamic virtual replicas of urban environments, enabling real-time monitoring, scenario simulation, and predictive analytics for city systems. Recent literature highlights their growing use in supporting decision and policy making, particularly for scenario modeling, sustainability planning, and adaptive governance. However, most implementations remain domain-specific and are not fully integrated with participatory governance or secure data management frameworks. For example, recent research demonstrates that integrating digital twin technologies with advanced data analytics at the neighborhood level significantly enhances urban resource management and resilience [10]. They provide a systematic review of digital twin applications in public sector decision support, emphasizing the need for context-specific models and integration with governance processes, especially in smaller or less digitally mature cities. The development of a Smart City platform based on digital twin technology enables real-time monitoring and simulation of urban processes, supports decision making, and enhances transparency and citizen engagement in city management, which are essential for sustainable urban development [67]. A growing body of literature now explores the convergence of digital twins, blockchain, and AI for Smart City management. We present an integrated platform that automates ETL processes, enables real-time analytics, and secures data integrity through blockchain-enabled audit trails. This approach is distinguished by a unified architecture by coupling digital twins, AI-driven analytics, and blockchain-based auditability across the urban data lifecycle [9]. Blockchain modules for recording citizen proposals, votes, and administrative actions, supporting transparent and auditable decision making. Open APIs and modular design for integration with legacy systems and adaptation to different city sizes and digital maturities. The comparative analysis across different approaches to Smart City digitalization is shown in

Table 1. Key features across different approaches.

Feature	Traditional Models	Recent Integrated Models	Proposed Platform
Digital Twin Scope	Domain-specific, siloed	Multi-domain, limited integration	Unified, real-time, cross-domain
Blockchain Use	Audit trails, transactions	Data provenance, some participatory tools	End-to-end integrity, participatory governance
AI Integration	Standalone analytics	Predictive analytics, not fully coupled	Embedded in ETL, scenario simulation
Data Interoperability	Low-moderate	Improving, but still challenging	High, via modular/open APIs
Citizen Engagement	Limited, portal-based	Some participatory pilots	Blockchain-enabled, auditable

, where are visible advances of our integrated approach.

The proposed cloud-based platform and business intelligence (BI) model integrates machine learning (ML) into the ETL process to automate data workflows, enhance data quality, and enable real-time analytics.

Table 2. Traditional ETL vs. proposed ML-enhanced ETL.

Feature	Traditional ETL	Proposed ML-Enhanced ETL
Transformation	Rule-based, manual scripting, static transformations.	ML-driven automation predicts optimal transformations, detect anomalies, and impute missing values dynamically.
Scalability	Limited to on-premise infrastructure; batch-oriented.	Cloud-native leverages modular architectures for elastic scaling.
Data Quality	Manual error detection, reactive fixes.	Proactive quality control due to ML models has anomaly data detection.
Processing Speed	Hours/days for large datasets.	Real-time capabilities enabled near-instantaneous processing of streaming data via AI/ML.
Cost Efficiency	High upfront infrastructure costs.	Cloud services reduce operational costs.

shows a detailed comparison of traditional and contemporary ETL methodologies. The proposed cloud-based Smart City platform, integrating machine learning-enhanced ETL processes, offers substantial advantages over traditional ETL and business intelligence systems. By leveraging ML-driven automation, the platform dynamically predicts optimal transformations, detects anomalies, and imputes missing values, resulting in more accurate and reliable data processing compared to rule-based, manual scripting used in conventional systems. This not only improves data quality but also achieves anomaly detection. It also enables real-time analytics, allowing for near-instantaneous processing of streaming data, a capability that is typically limited to batch-oriented, slower traditional ETL workflows.

Scalability is another key strength, while traditional ETL is constrained by on-premise infrastructure, the platform’s cloud-native, modular architecture allows for elastic scaling to accommodate growing data volumes and complex urban analytics. This reduces operational costs and supports continuous, automated updates and integration with diverse data sources. Overall, the platform significantly enhances operational efficiency, data integrity, and decision-making speed for both public enterprises and SMEs, fostering a more responsive and sustainable Smart City ecosystem

Although, despite significant progress, the literature identifies persistent challenges which our platform tries to solve.

• Data standardization and interoperability—Harmonizing data formats and integrating legacy systems remains a technical barrier [12].
• Privacy and compliance—Ensuring data protection and regulatory compliance, particularly with blockchain-based systems, is a critical concern [41].
• Scalability and generalizability—Most integrated platforms are still at the pilot stage, with further research needed to validate scalability in larger or more complex urban environments [9].

The integrated platform described in the article advances the state-of-the-art by uniting digital twin, blockchain, and AI technologies into a cohesive, scalable, and participatory Smart City solution. This approach addresses key limitations of traditional and current frameworks, as documented in recent scientific literature, and offers a replicable model for data-driven, transparent, and sustainable urban transformation.

4. Developing Digital Infrastructure for Smart Cities

4.1. Implementation of a Private Blockchain Network Using the Hyperledger Besu Platform

Implementation of a private blockchain network requires a thorough and multidisciplinary approach. Selecting the right platform, defining access rules, implementing security measures, and ongoing maintenance are key factors for successfully integrating blockchain technology into a Smart City platform. This approach not only improves data security and transparency but also allows organizations and companies to achieve additional value through automation and optimization of business processes.

We propose the deployment of an integrated digital twin, blockchain, and artificial intelligence (AI) platform in a Smart City, involving city administration, waste management companies, technology providers, and research institutions. This modular architecture enables the integration of additional city services (e.g., energy management and green mobility incentives) and supports open APIs for interoperability with other Smart City solutions. By supporting open APIs, the proposed platform is not only intended for Smart City companies and organizations, but also for all other companies and organizations that can share information with the platform. Using open APIs, information exchange with the city administration is enabled, allowing timely action by both sides. This approach fosters a collaborative and adaptive urban ecosystem where diverse stakeholders can securely interact, share data, and create innovative services, ultimately driving more responsive and efficient city management.

Blockchain technology has become a key component in the digital transformation of various industries, enabling security, decentralization, and transparency. However, private blockchain networks have developed in response to the needs of businesses and institutions that want to control access to their networks and ensure data privacy. Three prominent platforms can be used to build a private blockchain network, like Hyperledger Fabric, Hyperledger Besu, and Corda. Each of these platforms has unique characteristics and application areas, depending on business requirements, as shown in

Table 3. The comparison of various blockchain platforms.

Feature	Hyperledger Fabric	Hyperledger Besu	Corda
Blockchain type	Permissioned	Permissioned or Public	Permissioned (Peer-to-peer)
Architecture	Modular, Layered	Ethereum compatible	Notary network using DHT for peer-to-peer transactions
Consensus	Kafka, Raft, Solo, BFT variants	IBFT, QBFT, PoA, PoW	Notary service (Ensures transaction uniqueness and finality)
Smart contracts	Chaincode (Go, JavaScript, TypeScript)	Solidity (Ethereum EVM)	CorDapps (Kotlin, Java)
Transaction finalization	Very fast (no mining)	Depends on the consensus mechanism (slower with PoW)	Fast (transactions are final upon confirmation)
Privacy support	Private channels and Collection Policies	Private transactions are enabled via Tessera	Strictly enforced transaction privacy
Application area	Supply chain management, logistics, and healthcare	Enterprise Ethereum, fintech, banking	Banking, insurance, and financial services
Interoperability	supports interoperability with other blockchains	Compatible with Ethereum networks	Restricted interoperability with other blockchain networks

.

Hyperledger Fabric is a permissioned blockchain platform designed for business use, especially in industries that require high security, privacy, and scalability [56]. Its biggest advantage is its modular architecture, which enables flexible network configuration, selection of consensus protocols, and customization of data access rules. Fabric uses Chaincode, i.e., smart contracts, which can be written in Go, JavaScript, and TypeScript, which facilitates development and integration into existing business systems. One of the key features of Hyperledger Fabric is its fast finalization of transactions, because it does not use mining, but uses different types of ordering services like Kafka or Raft for consensus. In addition, it enables private channels and Collection Policies, which give users the option to make certain transactions visible only to selected participants. This platform is most often used in supply chains, logistics, healthcare, and digital identity, where it is necessary to precisely control who has access to certain data.

Hyperledger Besu is an EVM-compatible blockchain client, making it an excellent choice for enterprises that want to use private networks but retain the ability to connect to the Ethereum blockchain [59]. Unlike Hyperledger Fabric, which uses its smart contract system, Besu uses Solidity smart contracts, which allow development teams to port or reuse existing Ethereum applications. One of the key advantages is flexibility in the choice of consensus protocols, as it supports IBFT, QBFT, PoA (Proof of Authority), and PoW (Proof of Work) [43,68], which can be adapted to different application scenarios. Also, it enables private transactions through the Tessera system, which means that certain transactions can remain visible only within a defined group of participants. It is most often used in the financial sector, fintech applications, and private Ethereum networks, where it is important to maintain compatibility with the existing Ethereum ecosystem, but at the same time, enable security and privacy.

Corda is a permissioned blockchain platform designed primarily for the financial sector and intercompany transactions [57]. Unlike Hyperledger Fabric and Besu, which use a traditional blockchain model where transactions are recorded in blocks and distributed to all network participants, Corda uses a peer-to-peer model, where transactions are shared only between the involved parties. This approach enables a high level of privacy and reduces unnecessary network load. Corda uses smart contracts called CorDapps, which are developed in Java and Kotlin, which is an advantage for financial institutions that already use these technologies. Instead of a classic blockchain consensus, Corda uses a Notary service system, which validates transactions and prevents double spending, but does not require all transactions to be visible to the entire network. Due to its specific design, Corda is most commonly used in banking, insurance, and legal contracts, where it is essential that transactions are legally binding and private.

In this paper, we decided to use Hyperledger Besu as the private blockchain platform. The decision was based on several key advantages it offers, making it a suitable choice for enterprise blockchain solutions. It is an Ethereum-compatible client that combines the strengths of the public Ethereum network with the flexibility needed for permissioned enterprise use cases, including support for the Ethereum Virtual Machine (EVM) and Solidity smart contracts, allowing easy migration or integration with existing Ethereum-based applications and tools. Flexible consensus mechanisms support multiple consensus algorithms such as IBFT, QBFT, PoA, and PoW, allowing customization of performance, security, and decentralization trade-offs based on project requirements. It can operate in both public and private permissioned environments, offering versatility depending on the use case. Through integration with privacy solutions like Tessera, private transactions are enabled, which is critical for enterprise applications requiring confidentiality. Being part of the Ethereum ecosystem, Hyperledger Besu benefits from mature developer tools, active community support, and interoperability with a wide range of Ethereum-based services. Optimization for enterprise environments provides reliable performance with configurable consensus and network parameters, which enable easy communication and value exchange with existing public or consortium Ethereum blockchains.

While Hyperledger Fabric offers advanced modular architecture, comprehensive privacy controls, and high scalability, the choice of Hyperledger Besu was made due to its seamless integration with Ethereum standards, greater flexibility in consensus protocols, and the broad adoption of Ethereum technology in enterprise settings. This makes Besu a strategic choice for solutions aiming to leverage Ethereum-compatible infrastructure while maintaining permissioned blockchain capabilities.

For the purpose of testing the proposed blockchain model, a private network was implemented using Hyperledger Besu deployed across nine nodes. This architecture was designed to balance scalability, fault tolerance, and realistic testing conditions. Node Allocation and Network Topology are as follows:

• Validator Nodes: Four nodes operated as validators, participating in the consensus protocol. These nodes are responsible for proposing, validating, and finalizing blocks, ensuring the integrity and reliability of the network. Each validator can be managed by a distinct organization, supporting data partitioning and access control at the organizational level. The topology allows for seamless addition of new validator nodes as the network grows, enabling horizontal scalability.
• Non-Validator (Regular) Nodes: The remaining five nodes functioned as regular (non-validator) nodes. These nodes maintain a copy of the blockchain, submit transactions, interact with smart contracts, and provide Application Programming Interface (API) endpoints for client applications. While they do not participate in block validation, they ensure network robustness and facilitate connectivity for end users and applications.
• Privacy and Permissioning: Besu’s permission features were configured to restrict network participation to authorized nodes and accounts, ensuring a permissioned environment suitable for enterprise and Smart City use cases. Private transactions were enabled via integration with a privacy manager such as Tessera, allowing confidential data sharing between designated participants.

This modular and scalable architecture allows flexible adjustment of validator and regular node numbers according to network requirements. Hyperledger Besu’s support for multiple consensus protocols, robust permissioning, and privacy features makes it well-suited for enterprise-grade, permissioned blockchain deployments.

The network infrastructure was provisioned with nine workstations, each meeting or exceeding the minimum hardware specifications:

• CPU: Minimum four virtual CPUs
• Memory: At least 16 GB RAM
• Storage: SSD drives for low latency and high I/O throughput

Necessary software components installed on each node included

• Docker and Docker Compose for container orchestration and deployment consistency.
• Node.js runtime environment to support various network management tools and smart contract interaction.

The network was configured by generating cryptographic key pairs for each node and defining node identities and permissions using Besu’s permissioning configuration files. Communication between nodes was secured using TLS certificates, and privacy for sensitive transactions was enabled through integration with a privacy manager such as Tessera, ensuring that only authorized participants can access confidential transaction data. Nodes were deployed across the prepared infrastructure as four validator nodes, where each validator node was managed by a distinct organization, participating in the consensus protocol to propose, validate, and finalize blocks. This ensures data partitioning, access control, and network reliability. The other five non-validator nodes maintain a full copy of the blockchain, submit transactions, interact with smart contracts, and provide API endpoints for client applications. They do not participate in block validation but enhance network robustness and scalability.

The network topology allows for seamless addition of new validators or regular nodes as the number of participants grows, supporting horizontal scalability and flexible adaptation to changing requirements. Smart contracts can be developed in Solidity to automate specific business logic and workflows relevant to the test scenarios, leveraging Besu’s EVM compatibility. These contracts should be thoroughly tested in a controlled environment using industry-standard tools such as Remix, Truffle, or Hardhat before deployment to the live Besu network, ensuring correctness and robustness. Tests should validate the following:

• The integrity and reliability of peer-to-peer communication between nodes.
• Correctness and finality of transactions as processed by the consensus protocol.
• System performance under simulated transaction loads to evaluate stability and scalability.
• Fault injection scenarios to assess the network’s tolerance to node failures or malicious activity.

The network was launched by activating all nodes and verifying their connectivity within the defined peer-to-peer topology. Monitoring tools such as Prometheus, Grafana, or Splunk could be deployed to continuously track node health, transaction throughput, and latency metrics, thereby facilitating proactive network management and troubleshooting. Ongoing maintenance includes the periodic application of security patches to Hyperledger Besu binaries and container images, as well as regular updates to supporting software. Security best practices were enforced, including encryption of data at rest using disk-level encryption or encrypted storage volumes. Encryption of data in transit through TLS-secured communication channels between nodes. Regular audits of cryptographic keys and node permissioning configurations. To demonstrate the flexibility and cross-platform compatibility of the private blockchain network, the deployment utilized a heterogeneous environment:

• A subset of nodes was deployed on Windows OS inside Docker containers.
• Another subset operated on native Linux installations.
• The remaining nodes ran within Docker containers on Linux hosts.

This mixed deployment strategy showcased the ability of Hyperledger Besu to operate seamlessly across different operating systems and container orchestration environments, underlining the solution’s adaptability for various enterprise infrastructure setups.

4.2. Building the Foundation of a Smart City Digital Twin

To create a digital twin of a Smart City, it is necessary to collect a wide range of information and data. It is necessary to create the most accurate and functional digital replica of the physical urban environment. A digital twin of a city is not just a 2D/3D model, but a dynamic system that enables monitoring, simulation, and decision making in real time. Spatial and physical data about the city is needed, as well as historical and analytical data, and of course, real-time data from sensors and other IoT devices. This includes traffic data and environmental data (pollution and energy consumption), among others.

In this paper, we would like to present the beginnings of building a digital twin of a Smart City. A database was created to collect a large amount of data about the city of Osijek. All geographic locations of house numbers in the city of Osijek were first stored in the database. All locations are displayed in a geocoordinate system. To collect geo-locations in the city of Osijek, a web application has been created. The application uses different online services for collecting geo-locations by searching for parameters such as the name of the city, street name, and street number. The result is an .xml file with all geo-location data. To validate the collected data, geo-locations are plotted, resulting in the contour of the city of Osijek, as illustrated in Figure 4.

In Figure 5, we can compare the obtained coordinates with the actual map of the city.

These geolocations need to be entered into the proposed blockchain. For each location, which represents the house number of a real-world building, it is possible to create an identity, and later add additional possibilities to that identity. Such as invoices, construction permits, reporting of damage or malfunctions, etc., through smart contracts implemented in the blockchain network.

5. Evaluation and Future Work

The Smart City platform should be implemented in collaboration with a city administration, waste management companies, traffic companies, technology providers, research institutions, and all other interested parties. The deployment focused on three main domains: digital twin integration, blockchain-based participatory governance, and AI-driven Smart City management, as illustrated in Figure 6. Quantitative results from the implementation include AI-powered optimization of waste collection routes could reduce operational costs by approximately 15% and a decrease in fuel consumption by 12%, as measured over a three-month period in the selected urban district [69,70]. The authors also stated that smart waste bins equipped with IoT sensors enabled more accurate forecasting of waste generation, reducing unnecessary collections by 18%. Also, environmental impact is gained by improved scheduling and route optimization contributed to an estimated 10% reduction in CO₂ emissions from waste collection vehicles. According to the literature, the Smart City platform should also facilitate up to 9% increase in recycling rates, attributed to better monitoring and citizen engagement features. However, it should be noted that in the referenced case study [70]. The reduction in unnecessary collections was reported as a 15% decrease in collection frequency, while the increase in recyclables collected reached 15%.

The blockchain-enabled participatory governance module is designed to increase citizen proposals, participation, trust, and willingness to engage, which should be confirmed by post-pilot surveys. Qualitative feedback highlighted increased transparency and trust in municipal services due to immutable blockchain records, greater citizen empowerment, and enhanced cross-departmental collaboration within the city administration, facilitated by the integrated digital twin and open data dashboards. The digital twin component enabled real-time simulation of various urban scenarios, such as changes in waste generation patterns during public events or the impact of new recycling policies. Scenario modeling demonstrated that proactive adjustments to collection schedules could further reduce costs and improve service responsiveness, which we will try to achieve in our future work.

In our future work, the Smart City platform will be evaluated through a three-month pilot implementation in collaboration with the city administration and waste management companies. The evaluation will cover the domains of digital twin integration, blockchain-based participatory governance, and AI-driven Smart City management. Data will be collected from IoT-enabled waste bins, operational records of municipal services, blockchain transaction logs, and user participation modules. Key Performance Indicators (KPIs) will be continuously monitored and visualized on public dashboards, supporting data-driven decision making and continuous improvement, and are widely used in both academic research and real-world deployments to benchmark operational effectiveness, environmental impact, and stakeholder engagement [71]. The metrics will be standardized and KPIs defined in accordance with international frameworks such as the NIST Holistic KPI (H-KPI) Framework [71] and the digital twin evaluation framework [72]. Metrics included operational cost savings (%), reduction in fuel consumption (%), CO₂ emissions (tons/month), recycling rates (%), and citizen participation (number of proposals/votes). After the pilot, statistical analysis will be conducted using descriptive statistics and comparative analysis (e.g., before/after implementation), while qualitative feedback will be gathered through user surveys and interviews. Results will be presented using visual elements such as graphs, tables, and before/after diagrams to enhance accessibility and persuasiveness.

Technical Challenges to the proposed Smart City platform are integrating heterogeneous data sources (e.g., IoT devices, legacy systems, and citizen apps), which requires significant effort in data normalization and semantic harmonization. Inconsistent data standards and incomplete datasets occasionally limited the accuracy of analytics and simulations. It is necessary to ensure the platform’s scalability to support additional city services, and integration with external systems demanded a modular architecture and adherence to open standards. Retrofitting existing infrastructure with sensors can pose technical and logistical obstacles. To maintain robust security across all layers, particularly for blockchain-based digital identities and sensitive urban data, continuous monitoring, regular audits, and compliance with GDPR and other regulations are required [73,74].

Organizational and regulatory challenges require implementation and strong coordination among city departments, public enterprises, and private partners. Differing priorities and legacy workflows sometimes slowed decision-making and adoption. There is also talent and skills gap where is the need for specialized skills in AI, IoT, blockchain, and data analytics highlights within the city workforce, necessitating targeted training and capacity-building. Ambiguities regarding data ownership, access rights, and accountability frameworks complicated data sharing and platform governance are also present. Regulatory compliance, especially concerning privacy and data protection, is a persistent concern.

This paper shows that a Smart City integrated digital twin and blockchain platform can transform both public and private sector operations within the urban environment. By leveraging real-time data, advanced analytics, and participatory mechanisms, cities can achieve higher operational efficiency, regulatory compliance, and citizen trust. The modularity and open standards of the proposed architecture ensure scalability and alignment with EU strategies such as the Green Deal and Digital Europe Program.

5.1. Comprehensive User Survey

While the technical performance of the platform has been thoroughly evaluated, the platform is still in the development and pilot phase, and in future work, a comprehensive survey of end users will be conducted. Gathering structured feedback from these stakeholders is a key priority for the next phase of the project. A user survey is planned to assess satisfaction, usability, and perceived impact of the platform among different stakeholder groups. The survey will be designed to include both quantitative and qualitative questions, covering topics such as ease of use, perceived transparency, trust in data security, and willingness to participate in digital governance modules. The target groups will include citizens who interact with the platform (e.g., via participatory governance features), city staff involved in operations, and representatives from municipal enterprises.

The survey will be distributed online and in person, with invitations sent via city communication channels and through the platform itself. Data will be collected anonymously and analyzed using standard statistical methods to identify trends and areas for improvement. The results will be used to refine the platform and inform future development cycles. This approach will ensure that end-user perspectives are systematically incorporated into the evaluation process, providing a more holistic assessment of the platform’s effectiveness and societal impact.

5.2. Platform Adaptability and Modularity

The platform’s architecture is intentionally designed to support a wide range of urban environments, including smaller municipalities and cities with lower levels of digital maturity. This adaptability is achieved through several key features like modular implementation, interoperability and integration, efficiency, and cloud support. Cities can selectively implement platform components that align with their immediate needs and digital capabilities. For example, a smaller city may initially deploy only basic modules for data collection and waste management, and later expand to include advanced functionalities such as AI-driven analytics or blockchain-based participatory governance as resources and expertise develop. The use of open APIs and standardized data formats facilitates seamless integration with existing legacy systems and third-party solutions. This is particularly valuable for smaller cities, where IT infrastructure is often heterogeneous and fragmented. By lowering technical barriers, the platform enables phased and resource-efficient digital transformation. The platform supports cloud-based deployment and modular architecture, allowing cities with limited IT resources to leverage scalable solutions. This flexibility reduces the need for substantial upfront investment and makes the platform accessible to municipalities with constrained budgets or limited technical staff. Implementation plans include comprehensive training and technical support for local staff, as well as phased roll-out strategies and documentation. This ensures sustainable adoption, knowledge transfer, and the gradual development of digital competencies among city employees. Smaller cities can leverage national and EU funding opportunities aimed at supporting digital transformation and Smart City initiatives. The platform’s modularity allows for incremental investments aligned with available funding and strategic priorities.

Through modular design, interoperability, resource efficiency, and a focus on capacity building, the platform is well-suited for adaptation to smaller and less digitally mature cities. This ensures that the benefits of digital transformation and smart governance are accessible to a broad spectrum of urban communities, fostering inclusive and sustainable urban development. In future work, we plan to pilot the proposed model in a single city and later apply the given results in larger or more complex urban environments. Some KPIs, such as long-term environmental impact and sustained citizen engagement, require further longitudinal study. The initial investment in IoT infrastructure and system integration may be a significant challenge for smaller cities. We also plan platform scaling to additional Smart City services (e.g., energy management, water distribution) and enhancing interoperability with national and EU-level Smart City data ecosystems. By expanding participatory governance features, more diverse forms of citizen engagement are supported.

6. Conclusions

This work demonstrates that integrating digital twin technology, advanced machine learning, and a private blockchain network provides a robust and scalable foundation for Smart City transformation. The proposed platform enables real-time Smart City data management, transparent participatory governance, and AI-powered optimization of public services, resulting in measurable improvements in operational efficiency, environmental impact, and citizen engagement. Pilot implementation showed significant reductions in operational costs and emissions, as well as increased recycling rates and community participation. The modular, interoperable architecture supports both public enterprises and companies, ensuring adaptability to diverse urban environments and alignment with EU digital and sustainability strategies. While challenges remain in data integration, stakeholder coordination, and regulatory compliance, the proposed platform offers a replicable pathway for cities seeking to enhance transparency, trust, and sustainability through data-driven innovation. Future work will focus on scaling the platform to additional city services and expanding participatory features to further empower Smart City communities.

Despite the demonstrated benefits of the integrated Smart City platform, it is important to acknowledge several limitations of this study. First, the pilot implementation will be conducted in a single urban district and over a limited time period, which may affect the generalizability of the results to larger or more diverse urban environments. Second, challenges related to data integration, interoperability, and legacy system compatibility remain significant barriers to full-scale deployment. Third, the platform’s scalability and adaptability to different regulatory frameworks and organizational cultures require further validation in real-world settings. Finally, the current evaluation relies on a relatively small set of key performance indicators and may not fully capture all aspects of stakeholder satisfaction or long-term sustainability. Addressing these limitations in future research will be essential for ensuring the robustness and broader applicability of the proposed solution. As part of future work, during the project implementation, it is planned to conduct further studies and pilot implementations in larger and more complex urban environments to validate and extend the findings of this research.