Smart community edge platform providing stream content analysis, service migration, and service chaining (本文)
Year: 2020
Type: dissertation
Abstract: A smart community utilizes information technology to interconnect and manage community infrastructures.These networks consist of many Internet of Things(IoT) devices that provide different services to the end-users.In conventional networks, these sensor data send to cloud services for processing and management.However, cloud-based data processing introduces latency to the services.Fog computing techniques have been introduced to support these services at the network edge reducing the network latency.Smart community networks should support latency-sensitive services such as smart grid systems at the edge.In addition, Smart community services require service migration and service chaining to manage and distribute multiple services.For example, the current smart community edge(SCE) supports smart energy management services where data anonymization and data aggregation services should be chained, and the services should be migrated depending on the network's location and network traffic.SCE services can leverage generic hardware devices and network virtualization technologies to deploy the services without proprietary middleware devices.The current network virtualization methods mainly consider only core network applications.In contrast, smart community services operate on application layer data and process in-transit data to capture sensor data at the edge.Therefore, data extraction edge nodes that support sensor data processing are required to support these smart community services.A service-oriented container-based solution that processes data streams from sensors using conventional hardware will improve the applicability, compatibility, and latency of smart community services.To this end, a software-based edge node, namely, the SCE platform, was proposed to support smart community services.SCE supports data-tapping applications, especially for IoT devices, and has a stream processing feature with a comparatively shorter processing delay.This tapping and processing function on in-transit data was named stream content analysis(SCA).SCA captures in-transit data through zero copy stream reconstruction and string matching process.Afterward, SCE proposes a distributed rule application method to manage multiple services and distribute matched data to the services.SCE supports services through Docker containers to provide remote deployment, service migration, and service isolation.The real world SCE platform implementation allows SCE services to operate on 10Gbps links and apply 100 accumulated rules while maintaining less than 1ms latency using commodity hardware devices.To support SCE service migration, SCE proposes a consistently guaranteed migration method to support service migration to distribute the services depending on the nodes' availability.The proposed migration technique is designed to guarantee network consistency while migrating between nodes.Compared to existing container migration methods, the proposed migration reduces the migration data transfer through container layers and migrating only the streams affected by the migration application through SCA.The proposed container migration methods reduced the network downtime by more than 10% compared to conventional methods for containers with image sizes larger than 400MBs.Furthermore, SCE services require chaining to distribute sensor data efficiently to the edge nodes to apply multiple network services for a given traffic flow.To this end, SCE introduces a service function chaining-based request distribution method that utilizes proactive data collection and heuristics to analyze the network traffic and to select optimal SCE nodes.The SCE request distribution method reduces the end-to-end service latency by 10% compared to the available algorithms.The SCE platform provides commodity hardware-based SCA, distributed rule change application, service migration, and service chaining to support SCE services.At various stages throughout the Ph.D. process, I had interesting discussions, support, and valuable feedback from my supervisor Prof. Hiroaki Nishi.I would like to express my sincere gratitude to him for the valuable guidance and support throughout these years.I would like to give my
Language: English
Topic: IoT and Edge/Fog Computing
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7139680347