RAPTOR: um algoritmo ameta-heurístico para alocação de recursos em computação de borda veicular
Carregando...
Data
2023-08-11
Autores
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
A rápida evolução dos dispositivos móveis, seus aplicativos e a quantidade de dados que eles geram causaram um aumento significativo no consumo de largura de banda e também congestionamento no núcleo da rede. A Computação de Borda oferece uma solução parcial para as desvantagens de desempenho, com a extensão da nuvem até a borda da rede com ou sem a utilização de RSUs, que são estruturas com poder computacional, de armazenamento e comunicação. Com essa inspiração, a Computação de Borda Veicular surge para oferecer suporte a aplicativos veiculares. Esse novo paradigma depende de veículos como nós de extremidade para fornecer recursos de memória, armazenamento, processamento e rede para aplicativos de outros veículos com recursos limitados. Porém, essa alocação de recursos gera um problema adicional, os nós têm alta mobilidade pela rede e a capacidade de processamento está dispersa pela mesma. A partir daí, surge a necessidade de métodos eficientes que apoiem a tomada de decisão durante a alocação desses recursos, para que essa alocação seja feita em tempo real, sem atrasos e sem desperdício de poder computacional. Este trabalho propõe o \sigla, um algoritmo meta-heurístico para alocação de recursos em Computação de Borda Veicular. O algoritmo é baseado no método da Otimização do Falcão Harris e tem a finalidade de otimizar a escolha do melhor veiculo para ter seus recursos alocados à um serviço, com vista para a maximização de serviços atendidos e redução da subutilização de recursos. Também estudou-se os tipos de Computação de Borda e alguns métodos de otimização meta-heurísticos de soluções múltiplas, para tomada de decisão na alocação de recursos. A partir dos resultados, é possível observar que no cenário estabelecido (LuST) com TraCI real de mobilidade, o RAPTOR obteve melhores resultados que todos os seus concorrentes, com desempenho de 29% à 36% superior ao segundo colocado em duas das três classes de serviços estabelecidas ("Baixo" e "Médio"), considerando apenas os dados de "Serviços Alocados", isso nos três diferentes coeficientes de chegada de serviço. Para a última classe de serviços ("Alto"), essa margem foi um pouco mais discreta, de 6% à 14% superior ao segundo colocado.
The rapid evolution of mobile devices, their applications, and the amount of data they generate has caused a significant increase in bandwidth consumption and congestion at the core of the network. Edge Computing offers a partial solution to the performance disadvantages, with the extension of the cloud to the edge of the network with or without the use of RSUs, which are structures with computational, storage, and communication power. With that inspiration, Vehicular Edge Computing emerged to support vehicular applications. This new paradigm relies on vehicles as endpoints to provide memory, storage, processing, and network resources to resource-constrained applications on other vehicles. However, this allocation of resources generates an additional problem, the nodes have high mobility through the network and the processing capacity is dispersed throughout it. From there, the need arises for efficient methods that support decision-making during the allocation of these resources, so that this allocation is done in real time, without delays, and without wasting computational power. This work proposes the RAPTOR, a meta-heuristic algorithm for resource allocation in Vehicular Edge Computing. The algorithm is based on the Harris Hawk Optimization method and aims to optimize the choice of the best vehicle to have its resources allocated to a service, with a view to maximizing services served and reducing the underutilization of resources. We also studied the types of Edge Computing and some meta-heuristic optimization methods of multiple solutions, for decision making in resource allocation. From the results, it is possible to observe that in the established scenario (LuST) with real mobility TraCI, the RAPTOR obtained better results than all its competitors, with a performance of 29% to 36% superior to second place in two of the three classes of services established (“Low” and “Medium”), considering only the data of “Allocated Services”, this in the three different service arrival coefficients. For the last class of services (“High”), this margin was a little more discreet, from 6% to 14% higher than the second place.
The rapid evolution of mobile devices, their applications, and the amount of data they generate has caused a significant increase in bandwidth consumption and congestion at the core of the network. Edge Computing offers a partial solution to the performance disadvantages, with the extension of the cloud to the edge of the network with or without the use of RSUs, which are structures with computational, storage, and communication power. With that inspiration, Vehicular Edge Computing emerged to support vehicular applications. This new paradigm relies on vehicles as endpoints to provide memory, storage, processing, and network resources to resource-constrained applications on other vehicles. However, this allocation of resources generates an additional problem, the nodes have high mobility through the network and the processing capacity is dispersed throughout it. From there, the need arises for efficient methods that support decision-making during the allocation of these resources, so that this allocation is done in real time, without delays, and without wasting computational power. This work proposes the RAPTOR, a meta-heuristic algorithm for resource allocation in Vehicular Edge Computing. The algorithm is based on the Harris Hawk Optimization method and aims to optimize the choice of the best vehicle to have its resources allocated to a service, with a view to maximizing services served and reducing the underutilization of resources. We also studied the types of Edge Computing and some meta-heuristic optimization methods of multiple solutions, for decision making in resource allocation. From the results, it is possible to observe that in the established scenario (LuST) with real mobility TraCI, the RAPTOR obtained better results than all its competitors, with a performance of 29% to 36% superior to second place in two of the three classes of services established (“Low” and “Medium”), considering only the data of “Allocated Services”, this in the three different service arrival coefficients. For the last class of services (“High”), this margin was a little more discreet, from 6% to 14% higher than the second place.
Descrição
Palavras-chave
Ciência da computação, Redes ad hoc veiculares (Redes de computadores), Alocação de recursos, Algoritmos meta-heurísticos, Computer science, Vehicular ad hoc networks (Computer networks), Resource allocation, Meta-heuristic algorithms