Practical Aggregation in the Edge

• Autor(a) principal: Costa, Pedro Ákos Horváth Filipe da
• Data de Publicação: 2018
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10362/59502
• topic: Edge Computing; Wireless Ad Hoc Networks; Aggregation; Frameworks; Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Eletrotécnica, Eletrónica e Informática

Due to the increasing amounts of data produced by applications and devices, cloud infrastructures are becoming unable to timely process and provide answers back to users. This has led to the emergence of the edge computing paradigm that aims at moving computations closer to end user devices. Edge computing can be defined as performing computations outside the boundaries of cloud data centres. This however, can be materialised across very different scenarios considering the broad spectrum of devices that can be leveraged to perform computations in the edge. In this thesis, we focus on a concrete scenario of edge computing, that of multiple devices with wireless capabilities that collectively form a wireless ad hoc network to perform distributed computations. We aim at devising practical solutions for these scenarios however, there is a lack of tools to help us in achieving such goal. To address this first limitation we propose a novel framework, called Yggdrasil, that is specifically tailored to develop and execute distributed protocols over wireless ad hoc networks on commodity devices. As to enable distributed computations in such networks, we focus on the particular case of distributed data aggregation. In particular, we address a harder variant of this problem, that we dub distributed continuous aggregation, where input values used for the computation of the aggregation function may change over time, and propose a novel distributed continuous aggregation protocol, called MiRAge. We have implemented and validated both Yggdrasil and MiRAge through an extensive experimental evaluation using a test-bed composed of 24 Raspberry Pi’s. Our results show that Yggdrasil provides adequate abstractions and tools to implement and execute distributed protocols in wireless ad hoc settings. Our evaluation is also composed of a practical comparative study on distributed continuous aggregation protocols, that shows that MiRAge is more robust and achieves more precise aggregation results than competing state-of-the-art alternatives.