https://repositorio.ufu.br/handle/123456789/17903 2026-10-08T13:58:04Z https://repositorio.ufu.br/handle/123456789/50564 Title: Arquitetura reconfigurável (FPGA) baseada em modelo de Machine Learning para detecção de Ransomware em ambientes pervasivos Abstract: The growing number of ransomware attacks has put data security and service availabi lity at risk worldwide, causing an estimated financial impact of billions of dollars annually. Traditional signature-based detection approaches struggle to identify new variants due to obfuscation and polymorphism, limiting their effectiveness against zero-day attacks. At the same time, general-purpose processor architectures lack the hardware-level reconfi guration flexibility required for adaptive processing of Machine Learning (ML) models in pervasive computing environments, which are characterized by devices with limited processing, memory, and energy resources. This work proposes a ransomware detection architecture based on Field-Programmable Gate Array (FPGA) that combines Machine Learning techniques with the implementation of a classifier as combinational logic, priori tizing deterministic, low-latency inference for resource-constrained devices. The proposed method comprises training and comparing five supervised models (Decision Tree, Random Forest, eXtreme Gradient Boosting (XGBoost), Neural Network and Support Vector Ma chine (SVM)) on the CIC-MalMem-2022 dataset, using an 80/20 hold-out split combined with 5-fold cross-validation, followed by an explainability analysis using SHAP (SHa pley Additive exPlanations) to identify the consensus among models regarding the most relevant features for detection. The model that best balances predictive performance and structural simplicity will then be translated into a VHSIC Hardware Description Language (VHDL) description, in which the learned decision thresholds are embedded as constants in fixed-point comparators, enabling the synthesis of a single-cycle FPGA classifier. This work is expected to contribute to the state of the art in ransomware de tection by proposing a lightweight, low-latency hardware model suitable for embedded and pervasive devices with constrained computational resources, as an alternative to the high-performance, data-center-oriented architectures that predominate in the literature. 2026-08-10T00:00:00Z https://repositorio.ufu.br/handle/123456789/50412 Title: De voltas rápidas a decisões rápidas: uma arquitetura orientada a eventos para telemetria dos carros de Fórmula 1 Abstract: Formula 1 is a highly technological environment characterized by the continuous generation of large volumes of real-time telemetry data, which are fundamental for strategic and operational decision-making. In this context, the ability to collect, process, and make this information available with low latency becomes a key competitive advantage. This study proposes the development of a data architecture based on the event-driven paradigm, aiming to enable the simultaneous processing of real-time and batch data. The solution was designed to simulate the complete telemetry data flow of a Formula 1 car, using the F1 23 simulator as the source. The implemented architecture comprises a data ingestion module using the UDP protocol—responsible for capturing and pre-processing the information—followed by a messaging system that ensures scalable and fault-tolerant event transmission. From this layer, data is routed along two paths: a "fast layer" responsible for real-time processing and visualization, and a "batch layer" responsible for storage in a data lake and subsequent processing in a relational data warehouse. The adopted methodology is based on the construction and experimental validation of the pipeline using exclusively open-source technologies, focusing on performance, scalability, and component decoupling. The results demonstrate the viability of the proposed architecture for handling high-velocity and high-volume data, while also highlighting its potential for applications in real-world data engineering scenarios. It is concluded that an event-driven architecture is effective for processing real-time telemetry data, enabling not only instant analysis during the race but also structured storage for subsequent analysis and future applications, such as machine learning models. 2026-04-30T00:00:00Z https://repositorio.ufu.br/handle/123456789/50401 Title: ThreatLens: desenvolvimento de API para o back-end Abstract: This work presents the development of ThreatLens, a back-end application (REST API) responsible for centralizing, authenticating, and securely serving filtered and aggregated cyber threat intelligence messages already collected and classified by previous works developed at the School of Computer Science of the Federal University of Uberlândia (FACOM/UFU). To this end, a user authentication and management module was implemented based on JSON Web Token (JWT), with persisted, revocable refresh tokens, along with a module for querying messages (posts) collected from public Telegram channels, supporting filtering by source, relevance, category, and period, pagination, sorting, aggregated statistics, and word clouds. The multi-source data access layer was designed using the Strategy design pattern, allowing new sources to be incorporated without changes to the already implemented business logic. The functional evaluation, carried out through HTTP requests submitted to the implemented endpoints, confirmed compliance with the specified functional and non-functional requirements. As a result, the proposed architecture advances toward bridging the integration gap identified among the related works through a single, extensible API layer – although only the Telegram source is currently integrated, leaving the incorporation of new sources and integration with the alert module as directions for future work. 2026-08-26T00:00:00Z https://repositorio.ufu.br/handle/123456789/50382 Title: Detecção de intrusão em redes de computadores com aprendizado semi-supervisionado 2026-08-04T00:00:00Z