31Jul2026
09:00 Master's Defense By video conference
Topic on
Proposal and Evaluation of a Software Architecture for a Digital Twin of a Smart Campus
Student
Alvaro Martín Aspilcueta Narvaez
Advisor / Teacher
Juliana Freitag Borin
Brief summary
Digital Twins (DTs) have been gaining attention in both academia and industry due to their ability to replicate physical entities in virtual environments, enabling synchronization and transmission of data between the real and virtual domains. They offer cost reduction, risk mitigation, and process optimization by enabling simulations of hypothetical scenarios (what-if scenarios) and forecasts. This makes them suitable for enhancing services and optimizing resource utilization in Smart Cities.
However, urban-scale deployments face challenges related to infrastructure complexity and legal restrictions. In this context, a university's Smart Campus emerges as a promising environment for experimenting with projects that can later be scaled to an urban level. Intelligent Parking was selected as a use case due to the existence of a computer vision-based parking system in operation at the Institute of Computing, which can be evolved into a Digital Twin.
Several smart parking solutions have been proposed, using quite heterogeneous technologies, architectural decisions, and research objectives. Despite this, such solutions only partially address interoperability, considered one of the main obstacles to the adoption of digital technologies. Furthermore, the implementation of hypothetical scenario simulations is still underexplored, although it represents a fundamental characteristic of design dynamics. These aspects seem to be neglected due to the absence of a broader conception of digital design, which in some cases is reduced merely to a three-dimensional representation. Furthermore, the objectives of the research vary from user-oriented applications to system-oriented optimizations. The main gaps identified in Smart Parking solutions are related to interoperability, simulation of hypothetical scenarios, and the absence of experimental evaluations of distributed deployment strategies and scalability analysis of DT infrastructures.
This dissertation addresses these gaps through the proposition and experimental evaluation of a FIWARE-based Digital Twin architecture for Smart Parking, integrating NGSI-LD-based semantic interoperability, real-data-driven hypothetical scenario simulation, distributed deployment strategies for computer vision-based vehicle detection inference, and comprehensive operational analysis of the DT infrastructure across multiple computational paradigms such as mist, edge, fog, and cloud.
A factorial design was adopted to evaluate the impact on DT under different deployment strategies, workload levels, and traffic patterns. The experimental results were analyzed using parametric and non-parametric statistical approaches, such as three-way ANOVA and ANOVA based on Aligned Rank Transform (ART).
The results demonstrated that workload intensity and deployment architecture are the main factors affecting infrastructure behavior, while traffic variations have a limited operational impact. Furthermore, edge and fog deployments showed the best balance between scalability and operational stability, while CPU was identified as the main scalability constraint of the DT infrastructure.
Examination Board
Headlines:
| Juliana Freitag Borin | IC / UNICAMP |
| Breno Bernard Nicolau de França | IC / UNICAMP |
| Marcio Seiji Oyamada | UNIOESTE |
Substitutes:
| Bruno Barbieri de Pontes Cafeo | IC / UNICAMP |
| Daniel Macêdo Batista | IME / USP |