Navegando por Assunto "Vehicular networks"

Agora exibindo 1 - 2 de 2

Acesso aberto (Open Access)
Entropy-based client selection strategy for federated learning over vehicular network environment
(Universidade Federal do Pará, 2024-08-13) SOUSA, John Lucas Rodrigues Portilho de; ROSÁRIO, Denis Lima do; http://lattes.cnpq.br/8273198217435163; https://orcid.org/0000-0003-1119-2450; CERQUEIRA, Eduardo Coelho; http://lattes.cnpq.br/1028151705135221; https://orcid.org/0000-0003-2162-6523
Federated Learning (FL) emerges as a promising solution to enable collaborative model training for autonomous vehicles while preserving privacy and addressing communication overhead issues. Efficient client selection for participation in the training process remains challenging, especially in scenarios with statistical heterogeneity of data distribution and client failure events. Client failure, an uncontrollable event during training, reduces accuracy, convergence, and speed. This master thesis introduces an entropy-based client selection mechanisms for FL over Vehicular Network environments with client failure and non-IID data distributions. The proposed method is compared to a random selection mechanism in both IID and non-IID scenarios, as well as scenarios with random client drops. The results demonstrate that entropy-based selection outperforms other methods regarding training loss, accuracy, and Area Under the Curve (AUC), particularly in high client dropout and non-IID scenarios. These findings highlight the importance of considering entropy data for client selection to address the challenges posed by client failure and statistical heterogeneity in FL over Vehicular Network.
Acesso aberto (Open Access)
Mobility and cloud management in wireless heterogeneous 5g networks
(Universidade Federal do Pará, 2020-06-30) PACHECO, Lucas de Sousa; ROSÁRIO, Denis Lima do; http://lattes.cnpq.br/8273198217435163; HTTPS://ORCID.ORG/0000-0003-1119-2450; CERQUEIRA, Eduardo Coelho; http://lattes.cnpq.br/1028151705135221; https://orcid.org/0000-0003-2162-6523
The network mobility management branch is responsible for the protocols and actions taken by the network to ensure connectivity and the continuity of services consumed by mobile users. In this dissertation we analyse how next-generation networks pave the way for the distribution of video in vehicular networks (VANETs), composed by an heterogeneous ultra-dense infrastructure, joining existing wireless communication technologies to obtain greater spectral efficiency. A handover algorithm called HoVe is presented. Based on various criteria for video distribution on ultra-dense 5G VANETs. The simulation results show HoVe’s efficiency in providing videos with 19% higher quality than state-of-the-art algorithms, improving the package delivery rate by at least 30%. This work studies a particular case of VANETs that benefits from computing at the edge of the network, the case of Connected Autonomous Vehicles, or CAVs. Edge and mist computing are emerging solutions for remote data processing for autonomous vehicles, offering greater computational power, as well as the low latency required by autonomous driving. This work proposes the MOSAIC algorithm for service migration and resource management for communication between layers and between layers in edge and fog computing. Simulation results show the efficiency of the proposed algorithm with a better performance of up to 50% in terms of latency and five times less migration failures.