PPGSC - Mestrado em Sistemas e Computação

Permanent URI for this collectionhttps://repositorio.ufrn.br/handle/123456789/12059

Browse

Now showing 1 - 20 of 376

A migration metaprocess from a monolithic architecture to microservices
(Universidade Federal do Rio Grande do Norte, 2025-06-26) Medeiros, Henrique David de; Batista, Thais Vasconcelos; Cavalcante, Everton Ranielly de Sousa; https://orcid.org/0000-0002-2475-5075; http://lattes.cnpq.br/5065548216266121; https://orcid.org/0000-0003-3558-1450; http://lattes.cnpq.br/5521922960404236; http://lattes.cnpq.br/5136480770028616; Ferraz, Carlos André Guimarães
Microservice-based architectures have gained popularity for their ability to handle the complexity of cloud service-oriented systems and meet requirements for availability, maintainability, and scalability. However, the process of modernizing legacy systems to adopt microservices is often time-consuming and lacks a systematic approach. There is a need for clear guidance on which activities and artifacts are essential when migrating applications to microservices and how to establish a well-defined process to achieve the expected benefits. This work introduces the Metaprocess for Microservice Migration Kernel (M3K), which serves as a foundation for migrating monolithic applications to microservice architectures. The primary objective of this proposal is to establish a comprehensive metaprocess that development teams and organizations can use as a basis for defining their microservice migration processes. M3K leverages OMG’s Essence Standard, providing a solid framework for defining software development practices, activities, and work products, and facilitating the smooth transition to microservice architectures.
Desenvolvimento de modelos de previsão de consumo para Smart Meter
(Universidade Federal do Rio Grande do Norte, 2025-08-29) Sousa, Ewerton Leandro de; Kreutz, Márcio Eduardo; Cunha, Eduardo Nogueira; https://orcid.org/0000-0001-6991-7296; http://lattes.cnpq.br/7628513373242513; http://lattes.cnpq.br/6374279398246756; https://orcid.org/0009-0009-8188-6707; http://lattes.cnpq.br/1011490479268699; Pereira, Mônica Magalhães; http://lattes.cnpq.br/5777010848661813; Muller, Ivan; https://orcid.org/0000-0003-2914-0050; http://lattes.cnpq.br/1522479715721496
Increasing energy consumption around the world presents a number of significant challenges. Smart Grids play a crucial role in creating more sustainable and resilient energy systems. These systems are essential to face various energy and environmental challenges, allowing for more efficient management of resources, reduction of waste and integration of renewable energy sources. This research proposes the development of prediction models for Smart Meter, a smart meter capable of performing precise measurements, collecting consumption data in real time and executing prediction algorithms locally. The device integrates technologies for wireless communication and machine learning techniques optimized for edge devices. One of the main contributions of this work is the demonstration that it is possible to perform complex energy consumption predictions directly on the measuring device, reducing the need for continuous transmission of raw data and relieving the load on centralized systems. This not only improves the operational efficiency of smart grids, but also offers benefits in terms of consumers’ privacy and data security
Geração automática de questões de programação baseada em templates multicamadas
(Universidade Federal do Rio Grande do Norte, 2025-04-30) Silva, Abner de Santana; Aranha, Eduardo Henrique da Silva; http://lattes.cnpq.br/9520477461031645; Lucena, Márcia Jacyntha Nunes Rodrigues; Silva, Thiago Reis da
Developing programming skills in introductory courses requires frequent practice exercises, yet manually crafting questions is time-consuming and laborintensive. To address this challenge, this study investigates automatic question generation through multilayer templates supported by generative artificial intelligence, which suggests variations and provides feedback to students. A web-based tool was created that employs JSON files to define hierarchical templates and their question instances. A mixed-methods research design was adopted, including a case study with eight instructors to evaluate templateconstruction time, the number of questions generated, and perceptions gathered via questionnaires. Most instructors produced a complete template within 30 minutes, although they preferred adapting pre-existing models. The primary difficulties reported were the steep learning curve, direct JSON manipulation, and the absence of a graphical interface for visualizing layers. Future work involves expanding the template repository, developing a visual editor, and evaluating the solution with larger samples.
Utilizando aprendizado de máquina na identificação de Null pointer exceptions em análise estática de código em Java
(Universidade Federal do Rio Grande do Norte, 2023-01-30) Silva, Rodrigo Lafayette da; Cavalcante, Everton Ranielly de Sousa; Araújo, Daniel Sabino Amorim de; https://orcid.org/0000-0001-5572-0505; http://lattes.cnpq.br/4744754780165354; https://orcid.org/0000-0002-2475-5075; http://lattes.cnpq.br/5065548216266121; http://lattes.cnpq.br/3567167027912233; Abreu, Marjory Cristiany da Costa; https://orcid.org/0000-0001-7461-7570; http://lattes.cnpq.br/2234040548103596; Almeida, Rodrigo Bonifácio de; https://orcid.org/0000-0002-2380-2829; http://lattes.cnpq.br/0368311142108150
Mainstream object-oriented programming languages admit null values for references for the sake of flexibility. In Java, attempting to use an object reference with a null value throws a Null Pointer Exception (NPE), one of the most frequent causes of crashes in Java applications. Static analysis has been used to inspect the source or binary code to locate the origin of the exception by analyzing these artifacts without debugging-oriented program executions. Despite its effectiveness, static analysis relies on a fixed, static set of rules describing violation patterns, and it is known for a significant number of false positives. This study investigates how the use of Machine Learning (ML) techniques can improve the precision of detecting NPE-related faults through static analysis, a branch still unexplored in the literature and the software industry. The main goal is to propose, implement, and evaluate a classification-based approach to address the detection of NPErelated faults in Java code. The expected contributions from this work are: (i) understanding how ML techniques can be used to detect those faults via static analysis; (ii) providing a ML model to detect NPE-related faults; and (iii) an assessment of the performance of ML techniques in comparison to traditional static analysis tools. The results of the experiments showed that the new approach using Machine Learning (k-Nearest Neighbors) is more effective than traditional SATs, namely PMD, SpotBugs, SonarLint, and Infer, regarding NPE detection, presenting an average accuracy of 97.5% in its best configuration, albeit being up to 15 times less efficient in terms of relative performance.
Um método para geração automatizada de exercícios de demonstração com um nível de complexidade similar
(Universidade Federal do Rio Grande do Norte, 2025-07-21) Lopes Neto, João Mendes; Almeida, João Marcos de; Terrematte, Patrick César Alves; https://orcid.org/0000-0002-0385-0030; http://lattes.cnpq.br/4283045850342312; https://orcid.org/0000-0003-2601-8164; http://lattes.cnpq.br/3059324458238110; http://lattes.cnpq.br/2653822875045890; Santiago, Regivan Hugo Nunes; http://lattes.cnpq.br/7536988783793885; Olarte, Carlos; http://lattes.cnpq.br/1198550954813139; Seca Neto, Adolfo Gustavo Serra; https://orcid.org/0000-0002-0260-5922; http://lattes.cnpq.br/0071119715272492; Nalon, Cláudia; https://orcid.org/0000-0002-9792-5346; http://lattes.cnpq.br/7793795625581127
The automated generation of exercises may benefit educators by significantly reducing the time they spend in manually creating exercises. However, an obstacle in making such automation more present in an educator’s professional routine is controlling the level of complexity of mechanically generated exercises. In this work, we present a method for the automated generation of proof exercises with a comparable level of complexity. The inputs of this method are a proof exercise and a set of rules allowing to prove this exercise. The output is a set of proof exercises with a comparable complexity to that of given as input. The scope of exercises we work with are mathematical proof exercises described in first-order languages, covering topics such as Set Theory and Number Theory. In order to calculate the level of complexity of these exercises, we base our approach on the effort required to solve them via informal proofs. We argue that such an effort, in turn, may be captured by formal proofs in cut-based tableaux that do not contain logical symbols. The rules utilized in these proofs are extracted by a mechanizable procedure we provide. We use the analytic character of such rules and the formal structure tableau proofs have to provide a computational procedure of the method in question. As case studies, we demonstrate how our method works with fragments of Set Theory and Number Theory. A prototype implementation of the method was also developed and we present it here.
Arquitetura de microserviços para detecção e classificação de malária
(Universidade Federal do Rio Grande do Norte, 2024-07-26) Costa, Vanessa Dantas de Souto; Carvalho, Bruno Motta de; http://lattes.cnpq.br/0330924133337698; http://lattes.cnpq.br/1447567479085982; Canuto, Anne Magaly de Paula; Barroca Filho, Itamir de Morais; Codina, Daniel López
Malaria affects millions of people each year, predominantly in resourcelimited countries. According to the World Health Organization (WHO 2022), there were an estimated 619,000 Malaria deaths globally in 2021 compared to 625,000 in 2020. Thus, automated classification of Malaria-infected blood smear images is a critical component in improving the efficiency and accuracy of Malaria diagnosis. With the aim of enabling a flexible solution that would allow the integration of neural networks to treat diseases on a large scale, we propose a methodology for Malaria disease classification. Our approach involves data collection, preprocessing and YOLO object detection (in realtime), encapsulated into a microservices environment, creating a modular and scalable system that efficiently handles inference requests while ensuring flexibility and maintainability (as shown in the load test experiments). Thus, we are bridging the gap between fighting Malaria and technology and making an application that could serve as a blueprint for future works in the disease classification field.
Explorando Throttling para melhoria da disponibilidade em IoT dirigida à energia
(Universidade Federal do Rio Grande do Norte, 2025-03-26) Lopes, Paulo Henrique de Queiroz; Aquino Júnior, Gibeon Soares de; http://lattes.cnpq.br/1254338144161360; https://orcid.org/0000-0001-5514-0770; http://lattes.cnpq.br/2808561601422212; Delicato, Flávia Coimbra; Cacho, Nelio Alessandro Azevedo; Immich, Roger Kreutz
The expansion of the Internet of Things (IoT) requires addressing challenges related to energy supply, especially in environments with limited resources. Although batteries and supercapacitors can mitigate fluctuations, their continuous use raises environmental concerns. In this context, energy-driven systems propose an adaptive approach in which devices adjust their behavior according to variations in energy availability, enduring scarcity and leveraging energy abundance. Despite previous advances, there is still a lack of a structured approach that considers the implementation of the throttling mechanism adapted to the energy characteristics of the devices. In light of this, this work proposes the exploration of throttling as a mechanism to increase the availability of energy-driven IoT devices, in addition to defining a set of practical activities for its implementation. The methodology was validated through an experimental study with simulated devices, in which different throttling strategies were evaluated based on service rate and response rate to requested operations. The results showed that even in critical energy scenarios, throttling contributed to preserving the response capacity of the devices, outperforming uncontrolled approaches. It is concluded that the implementation of adaptive throttling, based on observable energy criteria, is a promising strategy to increase the availability of IoT devices in energy-constrained environments while promoting the appropriate use of available energy resources.
HARS1DE: arquitetura de hardware para processamento de CNNs1D na borda
(Universidade Federal do Rio Grande do Norte, 2025-01-31) Guimarães, Mailson Rodrigues de Medeiros; Kreutz, Márcio Eduardo; http://lattes.cnpq.br/6374279398246756; http://lattes.cnpq.br/3132811781302622; Zeferino, César Albenes; Fernandes, Marcelo Augusto Costa; Pereira, Monica Magalhães
There is a trend toward using the cloud computing paradigm, where resources, storage, and information processing are carried out in so-called "clouds"managed by providers. This paradigm is leveraged, for instance, to apply machine learning algorithms to large volumes of data. Conversely, there is the edge computing paradigm, where this processing load is transferred to elements closer to where the data is generated (at the network edge). Investment by technology companies in this type of computing and its techniques has been growing, as it can offer advantages, such as reduced processing latency, energy consumption, and resource demands that may not always be available in the cloud. Similarly to cloud computing, it is possible to apply predictive machine learning models at the edge, where hardware architectures dedicated to accelerating these processes can be employed. Thus, this work’s main objective is to implement, test, and validate a hardware architecture capable of accelerating the computation of 1D-CNNs inference, including pooling, activation, and dense layers, where performance metrics, accuracy, and hardware resource utilization are analyzed. Two representations of the architecture were developed to obtain the results: one in VHDL, synthesized for FPGA to get results regarding hardware resource allocation and timing, and another in Python, a high-level abstraction language, to obtain quicker results on the architecture’s behavior during longer processes, such as the computation of an entire neural network. Tests were conducted on three different variations of the proposed architecture.The results were obtained by applying the architecture in remote sensing, specifically for pixel classification in hyperspectral images. The neural network used was a simplified version of previous works to facilitate porting to hardware. In addition to being reconfigurable in the context of FPGAs, the resulting architecture exhibits adaptable behavior depending on the type of neural network layer being processed. Theoretical results demonstrate a maximum performance of 14.4GOP/s for the best architecture variation, as well as a maximum acceleration of 4.52× compared to an AMD EPYC 7B12 processor, 8.36× compared to an NVIDIA T4 and 3.39× to an AMD Ryzen 7 7800X3D. These results were achieved classifying one of the hyperspectral images and the best architecture variation ended with a FPGA resource usage below 80%.
Localização hipocentral: um estudo comparativo entre metaheurísticas
(Universidade Federal do Rio Grande do Norte, 2024-11-18) Silva, Arthur Anthony da Cunha Romão e; Carvalho, Bruno Motta de; Nascimento, Aderson Farias do; https://orcid.org/0000-0002-3961-5884; http://lattes.cnpq.br/8600906973888297; http://lattes.cnpq.br/0330924133337698; http://lattes.cnpq.br/2535790263841193; Santos, Edson José da Costa; Barboza, Francisco Márcio; Maia, Silvia Maria Diniz Monteiro
The determination of hypocentral location uses techniques to obtain the spatial coordinates of the hypocenter from temporal data observed at seismographic stations. This approach has applications ranging from geological characterization of the Earth’s crust to supporting incident prediction and hydrocarbon exploration. This work proposes the application of the OLC optimization algorithm and its hybridized version (OLCH), which incorporates genetic operators such as crossover, mutation, tournament, and elitism, as an alternative to traditional metaheuristics used in hypocentral inversion. Inspired by the behavior of grey wolf packs, the OLC, proposed by Seyedali Mirjalili in 2014, has been widely employed in optimization, system design, and machine learning. Its main mechanisms — hunting, encircling, and capturing — allow efficient exploration of the search space, finding solutions close to the global optimum in complex problems. The results were compared with algorithms widely used in hypocentral inversion, such as GN, as well as the hybridized version with MS, named GM, and other methods like AG and OEP. The comparison considered criteria such as objective function cost, computational time, and convergence. In tests with synthetic seismic data, OLC and OLCH significantly outperformed OEP in the objective function cost during the stabilization period, while OLCH surpassed GM in overall cost. With real data, OLCH stood out compared to OEP, both in the average objective function cost of the solutions and during the stabilization period. OLC also outperformed OEP, though only during the stabilization period. These results highlight the feasibility and advantages of OLC and OLCH in the context of hypocentral location using synthetic and real data, establishing them as innovative and alternative methodologies for hypocentral location.
Framework Guimarães: apoiando a construção de testes semi-automatizados para aplicações em Arduino
(Universidade Federal do Rio Grande do Norte, 2024-08-01) Negreiros, Sara Guimarães; Coelho, Roberta de Souza; Lucena, Márcia Jacyntha Nunes Rodrigues; http://lattes.cnpq.br/7982148962089986; http://lattes.cnpq.br/9854634275938452; http://lattes.cnpq.br/7082901769077209; Barbosa, Eiji Adachi Medeiros; Segundo, Francisco Carlos Gurgel da Silva
Arduino boards are used in embedded systems (ES) courses due to their low cost and support from the open-source community. This study investigated the importance of component and system testing in the context of teaching ES with the Arduino board. Through a survey with the Arduino community and interviews with educators, it was pointed out that tests are performed manually, which limits efficiency and makes it difficult to automate verification processes. As a result, the Guimarães framework was developed to perform semiautomated tests, both at component and system levels, integrating approaches such as Behavior Driven Development (BDD) and modeling with statecharts. In addition, a practical experience was conducted at the Instituto Metrópole Digital (IMD) where students developed and executed tests with the framework, integrating BDD practices. The students’ feedback highlighted the importance of these activities to ensure that the project meets the proposed requirements.
Vc-means: um novo algoritmo de agrupamento
(Universidade Federal do Rio Grande do Norte, 2024-08-09) Martins, Nicolas Jacobino; Bedregal, Benjamin René Callejas; http://lattes.cnpq.br/4601263005352005; http://lattes.cnpq.br/5588197462832398; Canuto, Anne Magaly de Paula; Silva, Huliane Medeiros da
This study presents the development and evaluation of the Vc-Means algorithm as an innovative approach to data clustering. The Vector c-Means (Vc-Means) is based on a previously developed algorithm called CK-Means and is designed to identify patterns and specific clusters in data sets. Statistical tests were conducted on 20 traditional data sets, comparing and validating its efficiency against three well-known algorithms in the literature: K-Means, Fuzzy C-Means (FCM), and Gustafson-Kessel (GK). The evaluation was performed using validation indices such as the DB index, Silhouette, Adjusted Rand Index, Calinski-Harabasz, Adjusted Mutual Information, and V-measure. The results showed that Vc-Means achieved great performance, with no significant statistical difference compared to the other algorithms, and demonstrated remarkable efficiency in terms of processing time.
Desenvolvimento e avaliação de um software para apoiar experimentos com sistemas interativos baseados em interface cérebro-computador
(Universidade Federal do Rio Grande do Norte, 2024-05-29) Lima, Franklin Matheus da Costa; Miranda, Leonardo Cunha de; https://orcid.org/0000-0003-1929-9391; http://lattes.cnpq.br/9064196799520278; https://orcid.org/0000-0002-8420-4865; http://lattes.cnpq.br/6918604953651306; Carvalho, Bruno Motta de; Pereira, Mônica Magalhães; Reis, Júlio César dos; Pereira, Roberto
Brain-Computer Interfaces (BCIs) enable interaction with computers through the user’s brain activity. In recent years, the BCIs have been gaining prominence due to technological advancements. In the context of BCIs, electroencephalography (EEG) is one of the methods used to read brain activity, and among the devices that are commonly used in this process, we can highlight NeuroSky’s MindWave headset. In this dissertation, a software was developed to support researchers in the development of studies related to the area of BCI, specifically those that make use of MindWave. This work describes, in detail, the entire context in which the software is used and the documentation needed to understand how it works. The software’s tools are discussed and all its features are presented, in order to show all the possibilities that can be achieved by using the software. Finally, the software is evaluated through a pilot study, which sought to verify its functionality in a BCI research scenario.
CLUPIR: um modelo para a classificação de linguagens de modelagem visual de software
(Universidade Federal do Rio Grande do Norte, 2024-06-07) Seabra, João Manuel Pimentel; Silva, Lyrene Fernandes da; https://orcid.org/0000-0003-1772-6062; http://lattes.cnpq.br/6082948468501249; http://lattes.cnpq.br/0283092030882917; Miranda, Leonardo Cunha de; Lucena, Marcia Jacyntha Nunes Rodrigues; Araújo, Camila de
Software Visual Modeling Languages (SVMLs) play a crucial role in facilitating systems analysis and documentation processes, as well as communication between those involved. However, the large number of languages available makes it difficult for the software designer to select an appropriate SVML to model a given problem situation. This research created a classification model (CLUPIR) that aims to organize and catalog SVMLs based on a set of aspects. The existing classification models and their classification aspects were surveyed through a systematic mapping of the literature, which supported the choice of the classification aspects of the CLUPIR model. At the end of the research, to demonstrate the use of the model, eight SVMLs were classified. After that, we validated the usefulness of the model with 30 professionals from the software industry.
Managing long-running asynchronous tasks in multi-tenant microservices
(Universidade Federal do Rio Grande do Norte, 2024-05-16) Silva, Felipe Morais da; Cavalcante, Everton Ranielly de Sousa; https://orcid.org/0000-0003-3558-1450; http://lattes.cnpq.br/5521922960404236; http://lattes.cnpq.br/7243232425776112; Cacho, Nélio Alessandro Azevedo; Ferraz, Carlos André Guimarães
A multi-tenant microservices architecture involving components with asynchronous interactions and batch jobs requires efficient strategies for managing asynchronous workloads. This work addresses this issue in the context of a leading company in the development of tax software solutions for many national and multinational companies in Brazil. A critical process provided by the company’s cloud-based solutions involves tax integration, which includes coordinating complex tax calculation tasks and needs to be supported by asynchronous operations using a message broker to ensure order correctness. These operations can be independent of each other, which characterizes the parallel process; or they can be dependent on each other, which characterizes the First in First out (FIFO) process. FIFO processes have additional restrictions regarding parallels. For this reason, we explored and implemented two approaches to managing asynchronous workloads related to tax integration within a multi-tenant microservices architecture in the enterprise context: (i) a polling-based approach that employs a queue as a Distributed Lock ( DL) and (ii) a push-based approach called Single Active Consumer (SAC) that relies on message broker logic to deliver messages. These approaches aim to achieve efficient resource allocation when dealing with an increasing number of container replicas and tenants. In this work, we evaluate the correctness and performance of DL and SAC approaches to clarify how asynchronous workloads impact the management of multi-tenant microservices architectures from a delivery and deployment perspective.
Implantação de inteligência artificial nativa em sistemas IoSGT: uma abordagem holística
(Universidade Federal do Rio Grande do Norte, 2024-04-30) Costa Filho, Paulo Eugênio da; Kreutz, Márcio Eduardo; Venâncio Neto, Augusto José; http://lattes.cnpq.br/1467664612924239; http://lattes.cnpq.br/6374279398246756; http://lattes.cnpq.br/8212965823478919; Rosário, Denis Lima do; Cunha, Eduardo Nogueira
The growing energy demand sharpens the quest for technological modernizations capable of meeting imminent needs, as well as increasing concern about mitigating the environmental impacts associated with this escalation. The state of the art in sg provides evidence of the use of IA techniques in IoSGT use cases, aiming to revolutionize the way energy is produced, transmitted, and consumed. Indeed, IA is presumed to offer unprecedented levels of disruption in the electrical sector through intelligent control methods that can unlock new value streams for consumers, while supporting a highly accurate, reliable, and resilient system. However, much research is still needed in this area, such as the positioning of IA based instances along the edge-cloud continuum, types of techniques and algorithms for each use case, efficient use of predictive analytics capable of forecasting future demands, fault detection, consumption patterns, and anomalies in the electrical grid that allow for proactive measures to enhance grid reliability, among many others. This research proposal aims to use energy consumption forecasting as a tool to optimize resource use, avoid waste, and contribute to environmental preservation, as well as to classify electronic devices to understand consumption patterns. This approach will be implemented through a holistic architecture called IAIoSGT (Artificial Intelligence native in IoSGT). IAIoSGT is designed to accelerate the adoption of IA and ML techniques in an approach that efficiently integrates data processing from the edge to the cloud. The evaluation of the IAIoSGT architecture, including its compliance, performance, and implementation feasibility, was conducted in two distinct testbeds, addressing both physical devices and Machine Learning algorithms. The first testbed focused on classifying and identifying electronic devices connected to the same electrical network, exploring ML algorithms such as KNN, SVM, MLP, NB, and DT; the second test focused on energy consumption prediction comparing the Naive and LSTM algorithms. These tests are essential to validate the effectiveness and robustness of the proposed approach, thus contributing to the advancement of the state of the art in IA applied to SG.
Técnicas de tolerância a falhas em perceptron multicamadas baseado em FPGA - estudo de caso: salve todas
(Universidade Federal do Rio Grande do Norte, 2024-05-31) Ribeiro, Maria Fernanda Cabral; Pereira, Mônica Magalhães; http://lattes.cnpq.br/5777010848661813; http://lattes.cnpq.br/3171119836039554; Canuto, Anne Magaly de Paula; http://lattes.cnpq.br/1357887401899097; Lopes, Alba Sandyra Bezerra; Kastensmindt, Fernanda Gusmão de Lima
The concept of fault tolerance can be understood as the ability of a system to maintain its correct operation even after the occurrence of a failure (Avizienis et al. 2004). This area of study emerged in the 1950s, aimed at dealing with failures in military and aerospace equipment operating in hostile and/or remote environments, and since then it has proven to be a prominent field of study, especially with the popularization of the use of computers and embedded systems. And that’s the research field of this work: the application of fault tolerance techniques in an Artificial Neural Network with Multilayer Perceptron (MLP) architecture embedded in an FPGA. The MLP is part of a system aimed at women’s safety that works on the identification of possible risk situations for users. The system has vital signs, sudden movements and geolocation sensors that provide information about the user’s current situation to the MLP Network responsible for analyzing these data. Since the MLP Network plays a critical role in identifying risk situations, it is necessary to apply techniques to increase the system’s reliability, aiming at greater safety for the user. Therefore, this work analyzes the gains and impacts of applying three fault tolerance techniques combined in embedded MLP. The techniques used include: refining the weights and biases of neurons in the network’s processing layers; changes in the MLP architecture, involving the removal of hidden neurons that are less sensitive to failures and the duplication of hidden neurons that are more sensitive to failures (a technique known as Augmentation); and the Triple Modular Redundancy of the neurons in the input and output layers of the network. The results obtained with the application of the three mentioned techniques contributed to significant gains in the overall reliability of the system. The advantages of applying the techniques combined stand out, thus maximizing improvements in Reliability for the system. Furthermore, it also draws attention to the advantages of applying the techniques for Refining the Weights and Biases of the MLP Network, Removal of Hidden Neurons that are less sensitive to failures, since these techniques do not add additional costs to the project and, in the case of the Removal technique, also brings improvements in processing and system latency.
Recuperação inteligente de desastres em sistemas de operação, gerenciamento e controle de infraestruturas 5G
(Universidade Federal do Rio Grande do Norte, 2024-04-30) Santos, Charles Hallan Fernandes dos; Venâncio Neto, Augusto José; http://lattes.cnpq.br/1467664612924239; http://lattes.cnpq.br/0099358331049231; Immich, Roger Kreutz; https://orcid.org/0000-0003-2483-6382; http://lattes.cnpq.br/0535777592588490; Fontes, Ramon dos Reis; Silva, Felipe Sampaio Dantas da
Management solutions for telecommunications ecosystems based on fifth-generation (5G) networks must be able to control infrastructure resources in a granular and automated manner. When adopting a management model, service providers must aim to meet the quality requirements of their customers. With this in mind, operators employ Operation, Management and Control (OMC) centers, an ecosystem that involves different technologies and tools that interoperate to provide operational functions designed to guarantee service level agreements (SLA) on an ongoing basis. In this sense, the unavailability of an OMC can represent a disaster for the service provider and its users, given the dependence of the devices on the telecommunications infrastructure. To ensure fault tolerance, disaster recovery systems (DRS) must operate on redundant instances of the OMC, where the backup unit must take over when its primary instance fails. A DRS can adopt proactive features that detect the disaster before it occurs, maximizing the availability of the OMC. In addition, the adoption of multiple backup instances has the potential to increase the recovery possibilities of a failed OMC. In this sense, a DRS must incorporate selection mechanisms to determine the best backup to take over operations. Considering a DRS with proactive failure detection, the selection algorithm must choose the backup OMC that performs the transition of its execution and its data before the disaster occurs, in order to increase its availability. In order to select the most appropriate backup OMC (i.e. candidate), it is necessary to carry out a survey of performance indices that determine whether a candidate is capable of transferring the main OMC within an estimated time for the failure to occur. Given the absence of similar mechanisms in the literature used in the context of disaster detection and recovery, this master’s research is dedicated to exploring Machine Learning techniques to develop a mechanism for selecting backup OMCs. In this context, ML is used to estimate the period required to migrate an OMC for each candidate, in the hypothesis of reducing or eliminating data losses by choosing the best backup OMC. To this end, iDRS (intelligent DRS) is introduced, which is based on intelligent mechanisms to act in the assignment of control of OMCs, in the hypothesis of provisioning a system available to maintain 5G networks throughout their useful life. iDRS collects multiple pieces of information that influence service migration to candidate locations, such as network performance metrics and virtualized infrastructure computing resources. From this, an ML algorithm categorizes the candidates in terms of migration time estimates. A case study based on an emulated testbed attests to the effectiveness of iDRS in terms of OMC data integrity compared to state-of-the-art algorithms.
Geotrack: an end-to-end framework to support moving objects
(Universidade Federal do Rio Grande do Norte, 2024-04-02) Silva, Alison Hedigliranes da; Cacho, Nélio Alessandro Azevedo; Lopes, Frederico Araújo da Silva; http://lattes.cnpq.br/9177823996895375; http://lattes.cnpq.br/4635320220484649; http://lattes.cnpq.br/4136158670357169; Aquino Júnior, Gibeon Soares de; Teixeira, Leopoldo Motta
The proliferation of intensive use of geolocation devices has led to considerable battery drain in these devices, typically stemming from excessive HTTP calls or extreme precision in device location, resulting in difficulties for applications requiring frequent location updates. Faced with this issue, this study presents a framework developed to collect, process, and visualize geolocation data on mobile devices, particularly smartphones, aiming to promote greater battery efficiency. The solution aims to optimize the development process of geolocation-focused applications through a Java framework, reduce battery consumption by automating HTTP call managementa and enhance geolocation data precision with an Android library, and provide data visualization capabilities with a JavaScript library.
EDRaft: um algoritmo de consenso orientado à energia
(Universidade Federal do Rio Grande do Norte, 2023-12-18) Barros, Glauber Mendes da Silva; Aquino Júnior, Gibeon Soares de; http://lattes.cnpq.br/1254338144161360; http://lattes.cnpq.br/3133071002168102; Delicato, Flávia Coimbra; Cacho, Nélio Alessandro Azevedo
A trend that computational systems have been following in recent years is the need to adapt to environmental issues such as sustainability and environmental preservation. As a result, an approach that has gained strength is energy harvesting, where devices collect energy from renewable sources in the environment, reducing or replacing the use of non-renewable energy and batteries. One challenge with renewable sources is their unpredictability, as the amount of collected energy depends on external events, with temporal and spatial variability. A concept that can be adopted to address this unpredictability is energy-aware computing. Energy-aware systems consider their energy needs, allowing them to adjust their behavior according to their energy situation at any given moment. One paradigm that has been growing over the years and is increasingly present in the population’s daily lives is the Internet of Things (IoT). However, because it is directly based on fundamentals from the field of Distributed Systems, it inherits problems and classical solutions from this area. One of them is the Byzantine consensus problem, which arises from the attempt to achieve agreement among elements that collaborate in a cluster. This work aims to propose EDRaft, an evolution of the Raft consensus algorithm based on the concepts of energy-aware computing. This evolution seeks to adapt its use in systems that harvest energy from the environment, addressing the unpredictability of this process.
Algebraization in quasi-Nelson logics
(Universidade Federal do Rio Grande do Norte, 2023-10-31) Lima Neto, Clodomir Silva; Rivieccio, Umberto; Almeida, João Marcos de; https://orcid.org/0000-0003-2601-8164; http://lattes.cnpq.br/3059324458238110; http://lattes.cnpq.br/0597230560325577; https://orcid.org/0000-0001-9835-9481; http://lattes.cnpq.br/6847191906266562; Santiago, Regivan Hugo Nunes; Biraben, Rodolfo Ertola
Quasi-Nelson logic is a recently introduced generalization of Nelson’s constructive logic with strong negation to a non-involutive setting. The present work proposes to study the logic of some fragments of quasi-Nelson logic, namely: pocrims (ℒQNP) and semihoops (ℒQNS); in addition to the logic of quasi-N4-lattices (ℒQN4). This is done by means of an axiomatization via a finite Hilbert-style calculus. The principal question which we will address is whether the algebraic semantics of a given fragment of quasi-Nelson logic (or class of quasi-N4-lattices) can be axiomatized by means of equations or quasi-equations. The mathematical tool used in this investigation will be the twist-algebra representation. Coming to the question of algebraizability, we recall that quasi-Nelson logic (as extensions of ℱℒew) is algebraizable in the sense of Blok and Pigozzi. Furthermore, we showed the algebraizability of ℒQNP, ℒQNS and ℒQN4, which is BP-algebraizable with the set of defining equations E(x) := {x = x → x} and the set of equivalence formulas ∆(x, y) := {x → y, y → x, ∼ x → ∼ y, ∼ y → ∼ x}.

SIGAA

Browse

collection.page.browse.recent.head