Browsing by Author "Wanzeller, Cristina"
Now showing 1 - 6 of 6
Results Per Page
Sort Options
- An IoT-Based Framework for Monitoring Photovoltaic BatteriesPublication . Sarmento, Gonçalo; Cardoso, Filipe; Mota, Mickael; Henriques, João; Abbasi, Maryam; Martins, Pedro; Wanzeller, Cristina; Caldeira, FilipeThe need for a more sustainable world is increasing the number of installations of photovoltaic systems including batteries as a way to postpone energy consumption when it is most needed. However, the chemical nature of batteries makes them unreliable and prompt to dangerous situations and can lead to a potential disaster, such as fires and explosions, in case the underline issues keep undetected for a long time. To overcome this scenario is a proposed framework for monitoring the temperature and voltage of batteries enabled by acquired data from Internet of Things (IoT) sensors. In case of these values become abnormal, a notification is triggered. The experimental results demonstrate the feasibility of the proposed framework to detect and notify hazard situations caused by battery faults.
- Comparing machine learning vs. humans for dietary assessmentPublication . Abbasi, Maryam; Cardoso, Filipe; Wanzeller, Cristina; Martins, PedroDue to the availability of large-scale datasets (e.g., ImageNet, UECFood) and the advancement of deep Convolutional Neural Networks (CNN), computer vision image recognition has evolved dramatically. Currently, there are three major methods for using CNN: starting from scratch, using a pre-trained network off the shelf, and performing unsupervised pre-training with supervised changes. When it comes to those with dietary restrictions, automatic food detection and assessment are critical.In this research, we show how to address detection difficulties by combining three CNNs. The different CNN architectures are then assessed. The amount of parameters in the examined CNN models ranges from 5,000 to 160 million, depending on the number of layers. Second, the various CNNs under consideration are assessed based on dataset sizes and physical image context. The results are assessed in terms of performance vs. training time vs. accuracy. Finally, the accuracy of CNNs is investigated and examined using human knowledge and classification from the human visual system (HVS). Finally, additional categorization techniques, such as bag-of-words, are considered to solve this problem.Based on the findings, it can be concluded that the HVS is more accurate when a data set comprises a wide range of variables. When the dataset is restricted to niche photos, the CNN outperforms the HVS.
- Comparison of Semi-structured Data on MSSQL and PostgreSQLPublication . Alves, Leandro; Cardoso, Filipe; Oliveira, Pedro; Rocha, Júlio; Wanzeller, Cristina; Martins, Pedro; Abbasi, MaryamThe present study intends to compare the performance of two Data Base Management Systems, specifically Microsoft SQL Server and PostgreSQL, focusing on data insertion, queries execution, and indexation. To simulate how Microsoft SQL Server performs with key-value oriented datasets we use a converted TPC-H lineitem table. The data set is explored in two different ways, firsts using the key-value-like format and second in JSON format. The same dataset is applied to PostgreSQL DBMS to analyse performance and compare both database engines. After testing the load process on both databases, performance metrics (execution times) are obtained and compared. Experimental results show that, in general, inserts are approximately twice times faster in Microsoft SQL Server because they are injected as plain text without any type of verification, while in PostgreSQL, loaded data includes a validating process, which delays the loading process. Moreover, we did additional indexation tests, from which we concluded that in general, data loading performance degrades. Regarding query performance in PostgreSQL, we conclude that with indexation, queries become three or four percent faster, and six times faster in Microsoft SQL Server.
- IoT-Based Monitoring System for Photovoltaic Battery ManagementPublication . Sarmento, Gonçalo; Cardoso, Filipe; Mota, Mickael; Henriques, João; Abbasi, Maryam; Martins, Pedro; Wanzeller, Cristina; Caldeira, FilipeAs number of photovoltaic systems being installed is increasing, also many users are deciding to include batteries, and postpone the consumption of that energy when it is most needed. Moreover,a crucial issue can keep undetected for long time causing dangerous situation which can lead to a potential disaster. However, the chemical nature of batteries makes them unreliable and dangerous. In this risky scenario, it becomes essential to monitor their behaviour in order to avoid accidents. In that purpose, a enabled Internet of Things (IoT) framework is proposed for monitoring the batteries values for temperature and voltage through the use of sensors. If these values becomes abnormal, a notification is triggered to the responsible person. Based on experimental results, the proposed framework to detect and notify hazard situations caused by battery faults.
- Performance Evaluation Between HarperDB, Mongo DB and PostgreSQLPublication . Figueiredo, Figueiredo; Cardoso, Filipe; Saraiva, Goncalo; Rebelo, João; Rodrigues, Ricardo; Wanzeller, Cristina; Martins, Pedro; Abbasi, MaryamSeveral modern-day problems, like information overload and big data, need to deal with large amounts of data. As such, to meet the application requirements, for instance, performance and consistency, more and more systems are adapting to the specificities. The existing Relational Database Management System (RDBMS)’s the processing of massive data has become an issue because these databases do not deal with a massive amount of data. NoSQL is a database management system that makes processing massive and/or unstructured data easier because it uses key-value to store the data, collections or document stores instead of tables. Many companies today tend to start a project using NoSQL. However, HarperDB aims to produce a relational and nonrelational DBMS, allowing developers to choose between different solutions. This paper aims to show the most relevant differences between HarperDB, MongoDB and PostgreSQL and compare their performances. Preliminary results show that PostgreSQL performs better with structured data, but HarperDB can integrate NoSQL and SQL, which can be a significant advantage to HarperDB compared to the other solutions.
- A scalable smart lighting framework to save energyPublication . Rebelo, João; Rodrigues, Ricardo; Henriques, João; Cardoso, Filipe; Wanzeller, Cristina; Caldeira, FilipeIn the past few decades, the urbanization area increased significantly, requiring enhanced services and applications to improve the lifestyle of its citizens. Lighting is one of the most relevant infrastructures due to its impact on modern societies, but it is also complex to manage them in cities since it involves a massive number of widespread posts and is costly as the result of the consumption of significant amounts of energy. In that regard, this work proposes a scalable framework to manage a significant huge number of lamp posts. Its purpose is to give support to collecting large amounts of sensor data to help to analyze and efficiently fit the light intensity level to the space the posts are covering. Luminosity sensors are used to optimize the intensity of light needed in the urban areas. The proposed framework explores the concept of smart cities by combining the data collected from sensors plugged into IoT (Internet-ofThings) devices. The proposed framework offers the capability to extend and integrate new services to different domains with each other which enhances the quality and performance of urban services. To demonstrate the feasibility of the framework, a simulation was put in place.