EPFIOT: Edge provisioning for IoT
This document describes the implementation of Epfiot (Edge Provisioning For Internet of Things), an appliance that follows an Infrastructure-as-a-service (IaaS) pattern targeting edge computing. Epfiot is specifically designed and implemented to interact with IoT devices, and to use ad-hoc hardware...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis de maestría |
| Fecha de publicación: | 2020 |
| País: | España |
| Institución: | Universidad Complutense de Madrid (UCM) |
| Repositorio: | Docta Complutense |
| Idioma: | inglés |
| OAI Identifier: | oai:docta.ucm.es:20.500.14352/9058 |
| Acceso en línea: | https://hdl.handle.net/20.500.14352/9058 |
| Access Level: | acceso abierto |
| Palabra clave: | 004(043.3) Internet of things Virtualization Infrastructure as a service IaaS Edge computing Cloud computing LWM2M LAN Local area network Appliance Router Virtual machine. Internet de las cosas Virtualización Infraestructura como servicio Computación en el borde Computación en la nube Red local Aplicación Enrutador Maquina virtual. Informática (Informática) 1203.17 Informática |
| Sumario: | This document describes the implementation of Epfiot (Edge Provisioning For Internet of Things), an appliance that follows an Infrastructure-as-a-service (IaaS) pattern targeting edge computing. Epfiot is specifically designed and implemented to interact with IoT devices, and to use ad-hoc hardware accelerators for specific purposes in the field of machine learning. Until the emergence of edge computing, cloud computing was the only choice for data processing derived from IoT. However, certain doubts arise about its strict necessity, and the requirement of sending all the data collected by the devices directly to Internet, paying for the (sometimes large) latency costs. Epfiot aims, with a contained use of resources, to offer a complete infrastructure stack to perform an integral processing of data obtained from devices in the local network, simplifying the way of providing virtual machines at the edge, through the use of specific accelerators for machine learning that perform inference on the data quickly, hence reducing the latency associated with the transfer to remote datacenters. The application exhibits a full GraphQL interface building an entire IoT ecosystem, using infrastructure on the edge thanks to Linux virtualization (KVM) and emerging technologies, such as LwM2M to provide device bootstrapping. |
|---|