Smart Cities Modeling: A Systematic Mapping

• The paper systematically maps 106 smart city studies (2011–2020) using strict inclusion criteria and keyword-based classification.
• Key methodologies include precise TRL assessment, dimensional categorization, and quantitative analyses like entropy-logistic modeling.
• Findings highlight practical implications for MDE tool support and the need for integrated frameworks to enhance stakeholder-centric urban planning.

The modeling of smart cities has evolved as a distinct interdisciplinary research domain, driven by the complexity, scalability, and societal impact of urban environments. Smart cities are conceptualized as urban systems characterized by automation, pervasive ICT integration, data-driven management, and multidimensional objectives such as sustainability, economy, governance, and citizen wellbeing. Systematic mapping studies have emerged as rigorous mechanisms for establishing the state-of-the-art in smart cities modeling, identifying research concentrations, gaps, and future directions, particularly through the lens of Model-Driven Engineering (MDE) and formal modeling paradigms (Rossi et al., 22 Aug 2025).

1. Research Objectives and Mapping Methodology

The systematic mapping study on smart cities modeling approaches sets out to survey, classify, and analyze the evolution, trends, and maturity of modeling research within the domain. Following the guidelines of Petersen et al., the study defines explicit research questions targeting: (a) the predominance of modeled smart-city dimensions, (b) target application fields, (c) formal modeling approaches, (d) solution maturity, (e) publication chronology, and (f) disciplinary context. The methodology encompasses database-driven identification (IEEE Xplore, Scopus, ACM Digital Library, dblp), keyword restriction to titles for context specificity, explicit inclusion and exclusion criteria (language, page length, focus), duplicate screening, and two-phase keywording/classification.

The refined corpus comprises 106 selected works spanning 2011–2020, each mapped by research type facet, TRL (Technology Readiness Level, 0–9), smart-city dimension, application field, modeling approach, and tool support. Each publication is assigned to one or more modeling categories via the function $M: P \rightarrow \mathcal{P}(A)$, where $P$ is the set of publications and $A$ the set of modeling approaches.

2. Dimensions and Application Fields in Smart City Modeling

Smart city research is categorized across seven dimensions following Boyd Cohen: Smart Mobility, Smart Environment, Smart Economy, Smart People, Smart Living, Smart Governance, Smart Things. The mapping study reveals "Smart Governance" as the most modeled dimension (35%), followed by "Smart Environment" (22%), with dimensions such as "Smart People," "Smart Mobility," "Smart Economy," "Smart Living," and "Smart Things" underrepresented.

Table: Smart‐City Dimension Distribution | Dimension | Count | Percentage | |--------------------|-------|------------| | Smart Governance | 37 | 34.9% | | Smart Environment | 23 | 21.7% | | Smart People | 13 | 12.3% | | Smart Mobility | 10 | 9.4% | | Smart Economy | 10 | 9.4% | | Smart Living | 8 | 7.5% | | Smart Things | 5 | 4.7% |

Application domains are clustered into "SC Domain" (design, management, administration, assessment), "ICT" (data management, services, communication), "Quality" (sustainability, security, privacy, safety, accessibility), and "Stakeholders" (participation, citizen-government communication). The largest concentration falls within "SC Domain" (47%), with "ICT," "Quality," and "Stakeholders" following (Rossi et al., 22 Aug 2025).

3. Taxonomy of Modeling Approaches

The mapping identifies twenty modeling approaches; the most prominent are business modeling (14.2%), architectural modeling (13.2%), and ontology-based modeling (11.3%). Data modeling, 3D modeling, UML/behavioral modeling, mathematical modeling, service-based modeling, KPI/evaluation modeling, and conceptual/DSLM approaches are also documented.

Table: Modeling Approach Distribution | Approach | Count | Percentage | |---------------------|-------|------------| | Business modeling | 15 | 14.2% | | Architectural | 14 | 13.2% | | Ontology | 12 | 11.3% | | Data modeling | 10 | 9.4% | | 3D modeling | 9 | 8.5% | | Behavioral/UML | 8 | 7.5% | | Mathematical | 7 | 6.6% | | Service-based | 6 | 5.7% | | Evaluation/KPI | 5 | 4.7% | | Conceptual/DSLM | 5 | 4.7% |

Business modeling (e.g., Business Model Canvas) frames organizational processes and value flows. Architectural modeling encompasses reference architectures, IoT, big-data, and microservices, structuring system components and communication. Ontological and data-modeling approaches formalize domain knowledge and entity relations using ER, LOD, and meta-modeling constructs. 3D models (CityGML, BIM) facilitate spatial, infrastructural visualization. Behavioral/UML, mathematical, and service-based models provide further abstraction and functional specificity across domains (Rossi et al., 22 Aug 2025).

4. Metrics, Simulation Algorithms, and Smart-Growth Modeling

A distinctive quantitative approach is proposed in "Modeling smart growth of cities through entropy and logistics" (Flamino et al., 2017): the smart growth metric employs weighted entropy to objectively aggregate city indicators across the "three E’s" (Economic Prosperity, Environmental Sustainability, Social Equity), standardizing raw indicators by Z-score and computing normalized proportions $f_{ij}$, indicator entropies $H_i$, and objective weights $w_i$. Sustainability indices $F_{EP,j}$, $F_{ES,j}$, $F_{SE,j}$ are formed as categorywise aggregates, then combined by distributive weights $\gamma_k$ to yield a scalar Growth Index $GI_j$:

$GI_j = \begin{bmatrix}\gamma_{EP}\\gamma_{ES}\\gamma_{SE}\end{bmatrix}^T \begin{bmatrix}F_{EP,j}\F_{ES,j}\F_{SE,j}\end{bmatrix}$

Growth forecasting leverages a differential logistic-weight framework wherein each Smart-Growth Initiative (SGI) contributes percentages $i_k$ to ten principles $p_k(t)$, their interactions modulated by weights $w_{kj}$. The system evolves via:

$$\frac{d p_k}{dt} = \beta_k i_k p_k \left(1 - \sum_{j=1}^{10} w_{kj} p_j\right)$$

where $\beta_k$ reflects initiative effectiveness. Forward simulation integrates these equations alongside Volterra-type models for sustainability, supporting ranking of SGIs by efficiency and effectiveness.

A plausible implication is that entropy-logistic approaches enable scalable objective assessment and prioritization of policy mixes, overcoming the subjectivity of traditional agent-based or system dynamics models, and permitting robust scenario comparisons.

5. Analysis of Research Type, Maturity, and Tool Support

The mapping study classifies papers by research type facet: solution proposals (54%), validation (24%), evaluation/field implementation (19%), and experience/opinion (3%). Technology readiness analysis confirms that none achieve TRL 9 (operational deployment); most reside at laboratory or prototype stages (TRL 2–4), few at pilot demonstration (TRL 7–8). Only 15 papers (14%) provide tool support, typically as web platforms for governance assessment or environmental monitoring; stakeholder-centric and IoT-mesh modeling tools are rare.

Temporal trends show modeling activity commencing in 2011, rising steadily post-2015, with conference publications predominant (62%) and interdisciplinarity evident in venue distribution across CS, social sciences, engineering, urban planning.

6. Comparative Evaluation and Hybrid Frameworks

Comparative analysis situates the entropy-logistic approach (Flamino et al., 2017) with agent-based, system-dynamics, and network models. Agent-based models excel at individual heterogeneity but lack scalability; system-dynamics models are deterministic but frequently overlook data-driven indicator weighting; network models emphasize flow efficiency in utilities, transport. The entropy–logistic hybrid uniquely integrates large-scale urban data weighting with nonlinear policy dynamics, allowing rapid simulation and robust SGI ranking.

A plausible implication is that integration of business, architectural, and ontological modeling with formal metrics (such as the Growth Index) may facilitate more generalizable, interoperable frameworks for smart city assessment and planning.

7. Future Directions and Impact on Model-Driven Engineering

Systematic gaps identified include underrepresentation of "Smart People," "Mobility," "Living," "Economy," and "Things," minimal stakeholder participation modeling, siloed paradigms, and limited tool maturity. Future directions emphasize:

• Development of dedicated formalisms for underrepresented dimensions.
• Stakeholder-centric modeling using visual DSLs and low-code platforms.
• Construction of reference hybrid modeling frameworks interlinking business, architectural, and ontology paradigms.
• Acceleration of maturity via MDE toolchains (DSLs, transformations, code generation), and industrial validation through pilot projects.
• Consolidation of a Smart City MDE Body of Knowledge, unifying standards and reusable modeling infrastructure.

For Model-Driven Engineering, there exists an opportunity to define SC-specific languages, enhance multi-paradigm tool support, and leverage infrastructure such as Eclipse Modeling Framework and Acceleo for comprehensive smart city platform generation (Rossi et al., 22 Aug 2025).

This suggests that systematic mapping studies provide essential roadmaps for both academic and industrial smart city modeling, guiding the transition toward operational, interoperable, and citizen-centric urban ecosystems.