Due to the limited computing resources of drones, it is difficult to handle computation-intensive tasks locally, hence, fog-based computation offloading has been widely adopted. The effectiveness of an offloading operation, however, is determined by its ability to infer where the execution of code/data represents less computational effort for the drone, so that, by deciding where to offload correctly, the device benefits. Thus, this paper proposes MonDroneFog, a novel fog-based architecture that supports image offloading, as well as monitoring and storing the performance metrics related to the drone, wireless network, and cloudlet. It takes advantage of the main machine-learning algorithms to provide offloading decisions with high levels of accuracy, F1, and G-mean. We evaluate the main classification algorithms under our database and the results show that Multi-Layer Perceptron (MLP) and Logistic Regression classifiers achieve 99.64% and 99.20% accuracy, respectively. Under these conditions, MonDrone-Fog works well in dense forests when weather conditions are favorable and can be useful as a support system for SAR missions by providing a shorter runtime for image operations.
Apesar dos avanços recentes que tornaram o uso de UAVs (Unmanned Air Vehicles) em diversos cenários uma realidade, há vários desafios a serem enfrentados para explorar todo o seu potencial. O uso de Fog Computing se torna uma saı́da para auxiliar estes dispositivos móveis com energia e hardware limitados a terceirizar toda a capacidade de processamento para um recurso remoto na borda da rede. Este trabalho apresenta a avaliação de desempenho da técnica de offloading com uso de drone e servidor cloudlet em ambiente de fog computing para a transmissão e processamento de imagens através de algoritmo tradicional de detecção e reconhecimento facial em tempo real.
levantamento de métricas para avaliação de desempenho da técnica de offloading com uso de drone e servidor Cloudlet em ambiente de Fog Computing para a transmissão e processamento de imagens através de algoritmo tradicional de detecção e reconhecimento facial em tempo real, em diferentes cenários, visando a análise do consumo energético do drone, ao longo do processo, a intensidade de sinal entre o drone e o orquestrador, e o tempo de processamento gasto em cada entidade, drone e maquinas virtuais.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.