A5102845861- Software-Defined Networks and 5G
- Advanced MIMO Systems Optimization
- Speech Recognition and Synthesis
- Millimeter-Wave Propagation and Modeling
- IoT and Edge/Fog Computing
- Speech and dialogue systems
- Speech and Audio Processing
- Explainable Artificial Intelligence (XAI)
- Vehicular Ad Hoc Networks (VANETs)
- Advanced Wireless Communication Techniques
- Music and Audio Processing
- Software System Performance and Reliability
- Phonetics and Phonology Research
- Retinal Imaging and Analysis
- Cooperative Communication and Network Coding
- Medicinal Plants and Neuroprotection
- UAV Applications and Optimization
- Protein Tyrosine Phosphatases
- Natural Language Processing Techniques
- IoT Networks and Protocols
- Health Promotion and Cardiovascular Prevention
- Multimedia Communication and Technology
- Software Testing and Debugging Techniques
- Telecommunications and Broadcasting Technologies
- Advanced Computing and Algorithms
Ericsson (Sweden)
2019-2024
Centro Hospitalar do Porto
2017-2022
Universidade do Porto
2022
University of Trás-os-Montes and Alto Douro
2022
Universidade Federal do Maranhão
2021
Universidade Federal do Pará
2010-2019
Instituto Federal de Educação, Ciência e Tecnologia do Pará
2018
The increasing complexity of configuring cellular networks suggests that machine learning (ML) can effectively improve 5G technologies. Deep has proven successful in ML tasks such as speech processing and computational vision, with a performance scales the amount available data. lack large datasets inhibits flourish deep applications wireless communications. This paper presents methodology combines vehicle traffic simulator ray-tracing simulator, to generate channel realizations representing...
The future mobile network has the complex mission of distributing available radio resources among various applications with different requirements. access slicing enables creation logical networks by isolating and using dedicated for each group applications. In this scenario, resource scheduling (RRS) is responsible slices to fulfill their service-level agreement (SLA) requirements, prioritizing critical while minimizing number intent violations. Moreover, ensuring that RRS can deal a high...
A espondilite anquilosante (EA) é uma doença inflamatória crônica do esqueleto axial que pode levar a deformidades estruturais e incapacidade funcional. O diagnóstico precoce continua sendo um desafio, vez os métodos tradicionais, como exames de imagem biomarcadores inflamatórios inespecíficos, apresentam limitações na detecção da em estágios iniciais. Este estudo teve objetivo revisar avanços recentes identificação aplicação para o prognóstico anquilosante. Foram analisadas evidências sobre...
The fifth-generation (5G) cellular networks incorporate a large variety of technologies in order to address very distinct use cases. Assessing these and investigating future alternatives is complicated when one relies only on simulators. 5G testbeds are an important alternative simulators many have been recently described, emphasizing aspects such as cloud functionalities, management orchestration. This work presents mobile network testbed with virtualized orchestrated structure using...
Artificial Intelligence (AI) is implemented in various applications of telecommunication domain, ranging from managing the network, controlling a specific hardware function, preventing failure, or troubleshooting problem till automating network slice management 5G. The greater levels autonomy increase need for explainability decisions made by AI so that humans can understand them (e.g. underlying data evidence and causal reasoning) consequently enabling trust. This paper presents first,...
Network slicing at the radio access network (RAN) domain, called RAN slicing, requires elasticity, efficient resource sharing, and customization. In this scenario, scheduling (RRS) is responsible for dealing with scarce limited frequency spectrum resources available domain while fulfilling slice intents. The wide variety of scenarios supported in 5G beyond networks makes RRS problem scenario a significant challenge. This paper proposes an intent-aware reinforcement learning method to perform...
Glaucoma is a silent disease characterized by progressive degeneration of retinal ganglion cells and, when not detected or treated early, can lead to blindness. Computer systems have demonstrated their efficiency in the medical decision-making process and Artificial Intelligence (AI) techniques helped advances ophthalmology, allowing for faster more effective detection glaucoma. Machine learning very promising subfield AI that supports research understanding development, progression...
Fifth-generation (5G) cellular communication systems have embraced massive multiple-input-multiple-output (MIMO) in the low- and mid-band frequencies. In a multiband system, base station can serve different users each band, while user equipment operate only single band simultaneously. This paper considers MIMO system where channels are dynamically allocated frequency bands. We treat as scheduling resource allocation problem propose deep reinforcement learning (DRL) agents to perform...
This paper describes a framework for research on Reinforcement Learning (RL) applied to scheduling and MIMO beam selection. consists of asking the RL agent schedule user then choose index beamforming codebook serve it. A key aspect this problem is that simulation communication system artificial intelligence engine based virtual world created with AirSim Unreal Engine. These components enable so-called CAVIAR methodology, which leads highly realistic 3D scenarios. modeling adopted in also...
There are several commercial text-to-speech (TTS) systems that generate speech signals sound very natural. A distinct problem is utterance copy, which consists in taking as input (instead of text, TTS) and find the parameters would drive a synthesizer to mimics target with respect contents speaker identity. Utterance copy difficult task due need adjusting their nonlinear relation output. Genetic algorithms (GA) have been used this embedded an analysis-by-synthesis loop, requires solving...
Purpose: To characterize the population of diabetics referred to Ophthalmology Department Centro Hospitalar do Porto (CHP) from screening program ARS Norte, evaluate this type method and perceive its impact in dynamic an Department. Methods: Retrospective evaluation clinical processes diabetic patients CHP Diabetic Retinopathy (DR) Screening Consultation Norte program, between January 2012 December 2016. The variables analyzed [...]
Digital twins are an important technology for advancing mobile communications, specially in use cases that require simultaneously simulating the wireless channel, 3D scenes and machine learning. Aiming at providing a solution to this demand, work describes modular co-simulation methodology called CAVIAR. Here, CAVIAR is upgraded support message passing library enable virtual counterpart of digital twin system using different 6G-related simulators. The main contributions detailed description...
A main challenge of using 5G network slices is to meet all the quality service requirements (which are agreed with customer in a level agreement (SLA)), throughout slices' lifecycle. To avoid penalty for violation, proactive solution presented, including predicting SLA explaining violation cause, and then providing an adaptation traffic. This work uses counterfactual (CF) explanations 1) explain factors affecting identified model's prediction 2) present modifications input values, which...
Machine learning has become a powerful tool for improving vehicle-to-vehicle (V2V) communication systems, and in general requires large datasets model training assessment. However, creating realistic using field measurements is challenging due to the bandwidths involved usage of multiple antennas. Simulations have been widely adopted circumvent relative high cost measurement campaigns. This paper presents development new public dataset research within V2V scenarios, machine algorithms that...
Utterance copy, also known as speech imitation, is the task of estimating parameters an input, target signal in order to artificially reconstruct another with same properties at output. This can be considered a difficult inverse problem, since input-output relationship often non-linear, apart from having several estimated and adjusted. work describes development application that uses long short-term memory neural network (LSTM) learn how estimate input thel formant-based Klatt synthesizer....