Timely Information Updates for the Internet of Things with Serverless Computing
Timely Information Updates for the Internet of Things with Serverless Computing
复制标题
DOI：
10.1109/icc42927.2021.9500618
复制
发表时间：
2021-06
期刊：
影响因子：
--
通讯作者：
Sonori Wakisaka;Yi-Han Chiang;Hai-Hsing Lin;Yusheng Ji
中科院分区：
文献类型：
--
作者：
Sonori Wakisaka;Yi-Han Chiang;Hai-Hsing Lin;Yusheng Ji
近年来，物联网（IoT）应用的激增引起了越来越多的研究关注。鉴于各种物联网应用对延迟的敏感性，信息年龄（AoI）被广泛认为是一种有前途的性能指标，用于量化物联网设备信息更新的及时性（即新鲜度或年龄）。此外，无服务器计算（也称为功能即服务 (FaaS)）发展成为一种高度可扩展且灵活的计算架构，可以促进及时的物联网分析。事实上，无服务器功能的执行可能依赖于从物联网设备收集的数据，因此物联网设备信息更新的新鲜度会对无服务器功能的服务年龄（AoS）产生直接影响。尽管现有的工作致力于无服务器计算的各个方面，但很少研究信息更新如何影响无服务器功能的 AoS 的问题。在本文中，我们通过无服务器计算解决物联网的信息更新传递和获取（IUDA）问题，并将其制定为整数线性程序（ILP），其目标是最小化无服务器函数的 AoS 加权和。为了解决 IUDA 问题，我们提出了一种离线和在线算法，分别在不知道无服务器功能到达的情况下调度信息更新。我们的仿真结果表明，所提出的解决方案在 AoS 性能方面优于现有解决方案，并有效地提供无服务器功能，并在各种参数设置下及时更新信息。
The proliferation of the Internet of Things (IoT) applications has resulted in the ever-increasing research attention in recent years. In light of the sensitivity to latency in various IoT applications, the age of information (AoI) has been widely regarded as a promising performance metric to quantify the timeliness (i.e., freshness or age) of information updates from IoT devices. In addition, serverless computing (also known as function as a service (FaaS)) evolves as a highly scalable and flexible computing architecture that can facilitate timely IoT analytics. In fact, the execution of serverless functions may rely on the data collected from IoT devices, therefore the freshness of information updates of IoT devices has prompt impacts on the age of service (AoS) of serverless functions. Although existing works have been devoted to various aspects of serverless computing, the issues regarding how information updates affect the AoS of serverless functions are rarely investigated. In this paper, we address the information update delivery and acquisition (IUDA) problem for IoT with serverless computing and formulate it an integer linear program (ILP), the objective of which is to minimize a weighted sum of AoS of serverless functions. To cope with the IUDA problem, we propose an offline and an online algorithm for scheduling information updates with and without the knowledge of the arrivals of serverless functions, respectively. Our simulation results demonstrate that the proposed solutions outperform existing solutions in terms of the AoS performance and effectively provide serverless functions with timely information updates under various parameter settings.