Littérature scientifique sur le sujet « MOBFOGSIM »

The object of research is an approach of functionality extension for simulation toolkits based on iFogSim. It is assumed by the native approach that enhancement of functionalities should be achieved by inheriting the fog device class and defining new features in its body. However, this approach makes it impossible to use inherited simulators together and significantly decreases flexibility even when utilizing a single simulator. Another problem related exclusively to iFogSim is a specific communication scheme between application modules, which results in data routing limitations in fog architectures and odd data streams taken into account. This paper introduces an alternative extension approach incorporating a peculiar inheritance scheme which tries to reconsider the standard approach from a behavioral design patterns point of view. The key feature of the suggested approach is an extraction of fog device features from the native class into separate behavioral classes. Meanwhile, the designed inheritance scheme allows to flexibly override and combine behaviors. According to the approach principles the developed simulator extends iFogSim with application modules addressing capabilities solving limitations, along with implementing users’ mobility and dynamic wireless connectivity as it is done in MobFogSim. With the aim to check its correctness, the designed toolkit was validated with the standard for iFogSim case study of «EEG Tractor Beam game» application. The validation included four scenarios. In the first two scenarios the features of users’ mobility and dynamic base station connectivity were validated. And in the next scenarios that utilized address routing the obtained delay and network usage values were compared with theoretically calculated ones. The validation results indicated the correct simulator behavior, and introduced functionalities extension approach, being more complex in comparison with the inative one, can significantly improve flexibility of the simulator

Fog Computing (FC) extends the Cloud towards the network edge. It provides end devices with access to resources and services that are located in topological proximity to them. This proximity enables key benefits (e.g., low latencies, reduced bandwidth consumption) that are not achievable when relying on Cloud-only solutions. FC is a horizontal paradigm in the sense that it is generic enough to be leveraged in a number of different application domains. However, its characteristics make it particularly suitable for the Internet of Things (IoT). After providing a survey on FC for the IoT, this thesis focuses on a specific research problem introduced with FC: the issue raised by device mobility in a FC environment. When a (IoT) device moves, the topological distance to the current Fog node (i.e., the Fog resource hosting the Fog service) may increase. Therefore, device mobility may impair the FC benefits, which are a result of Fog proximity. The objective is to support device mobility, namely to provide the FC benefits even in the presence of mobile devices. The most popular approach in literature to achieve this purpose is by migrating the Fog service across the Fog infrastructure, thus to let it be always close enough to the served mobile device. We refer to this paradigm as Companion Fog Computing (CFC), an extension of standard FC where the Fog service behaves as a "companion" of the mobile application. In this thesis, we first analyse the different aspects that characterise CFC, by reporting the state of the art for each aspect and highlighting the open issues and research opportunities. Then, we propose our own solutions. Specifically, we first consider Fog services as application containers and set up a small-scale FC testbed to perform a quantitative evaluation and comparison of the existing container migration techniques. The objective of this work is twofold: (i) clarify whether there exists a technique that performs the best under any condition or, otherwise, understand which technique is the most appropriate under which condition; (ii) better understand the inner workings of container migration. The second contribution is the proposal and validation of a platform, which we call Companion Fog Platform(CFP), that provides the necessary mechanisms to support device mobility in the Fog. Our CFP implements Fog services as containers and migrates them between Fog nodes according to the state-of-the-art container migration techniques. The third work described in this thesis is MobFogSim, a simulator that extends iFogSim to model device mobility and service migration in FC. After providing its design and implementation details, we validate the simulator. This is done by reproducing in MobFogSim the same conditions under which we evaluated the container migration techniques and by comparing the simulation results against those over the real testbed.

O fatiamento de rede tem se apresentado como uma tecnologia promissora para o gerenciamento de recursos em redes de computadores modernas como o 5G, criando diferentes redes virtuais sobre uma mesma infraestrutura física. Considerando escalabilidade, flexibilidade e aspectos financeiros, simuladores têm se tornado um possível ambiente para avaliar novas soluções desenvolvidas para esse cenário. Nesse contexto, o simulador MobFogSim foi desenvolvido para avaliar soluções em ambientes de Computação em Nuvem e Névoa, incluindo o suporte para migração de serviços, fatiamento de rede e gerenciamento da mobilidade dos usuários. Neste trabalho, apresentamos as novas funcionalidades do simulador em termos de escalabilidade e suporte ao fatiamento ponta-a-ponta, o que inclui recursos de armazenamento e processamento. Simulações considerando dados reais como parâmetros mostraram o impacto de diferentes critérios na alocação de recursos das redes virtuais, além do aumento da escalabilidade do simulador.

A Computação em Névoa é responsável por prover recursos computacionais na borda da rede a usuários com diferentes características e demandas. Por meio da virtualização de seus recursos, é possível criar múltiplas Redes Virtuais, ou Fatias de Rede, sobre uma mesma arquitetura física, cada uma atendendo um grupo de usuários. Nesse contexto, avaliar o desempenho de tais redes em diferentes cenários se torna primordial para identificar pontos fortes e fracos a serem considerados no desenvolvimento de mecanismos para gerenciar a rede. Este artigo apresenta uma análise de desempenho de diferentes abordagens de alocação de Fatias de Rede com o objetivo de otimizar o processo de migração de serviços na névoa. Resultados obtidos no simulador MobFogSim apontam que, devido a variações de demanda, o desempenho da alocação estática dessas redes pode se degradar ao longo do tempo. A alocação dinâmica de fatias de rede se apresentou como uma solução para esse cenário, a depender do custo computacional necessário para a reconfiguração dessas redes.