Communications Department, Universitat Politècnica de València (UPV), 46022 Valencia, Spain.
Sensors (Basel). 2023 Mar 8;23(6):2931. doi: 10.3390/s23062931.
Most recent edge and fog computing architectures aim at pushing cloud-native traits at the edge of the network, reducing latency, power consumption, and network overhead, allowing operations to be performed close to data sources. To manage these architectures in an autonomous way, systems that materialize in specific computing nodes must deploy self-* capabilities minimizing human intervention across the continuum of computing equipment. Nowadays, a systematic classification of such capabilities is missing, as well as an analysis on how those can be implemented. For a system owner in a continuum deployment, there is not a main reference publication to consult to determine what capabilities do exist and which are the sources to rely on. In this article, a literature review is conducted to analyze the self-* capabilities needed to achieve a self-* equipped nature in truly autonomous systems. The article aims to shed light on a potential uniting taxonomy in this heterogeneous field. In addition, the results provided include conclusions on why those aspects are too heterogeneously tackled, depend hugely on specific cases, and shed light on why there is not a clear reference architecture to guide on the matter of which traits to equip the nodes with.
大多数最新的边缘和雾计算架构旨在将云原生特性推向网络边缘，减少延迟、功耗和网络开销，允许在靠近数据源的地方进行操作。为了以自主的方式管理这些架构，在特定计算节点中实现的系统必须部署自我功能，最大限度地减少整个计算设备连续体中的人为干预。如今，缺少对这些功能的系统分类，以及关于如何实现这些功能的分析。对于连续部署中的系统所有者来说，没有主要的参考出版物可以咨询，以确定存在哪些功能以及依赖哪些来源。在本文中，进行了文献综述，以分析实现真正自主系统的自我装备性质所需的自我*功能。本文旨在阐明这一异构领域中潜在的统一分类法。此外，提供的结果还包括关于为什么这些方面处理得过于异构、在很大程度上取决于特定情况以及为什么没有明确的参考架构来指导节点配备哪些特性的结论。
