An exchange-based AIoT platform for fast AI application development

Liang, Yu-Cheng, Wu, Kun-Ru, Tong, Kit-Lun, Ren, Yi ORCID: https://orcid.org/0000-0001-7423-6719 and Tseng, Yu-Chee (2023) An exchange-based AIoT platform for fast AI application development. In: Proceedings of the 19th ACM International Symposium on QoS and Security for Wireless and Mobile Networks. Association for Computing Machinery (ACM), pp. 105-114.

[thumbnail of 3616391.3622770]
Preview
PDF (3616391.3622770) - Published Version
Download (2MB) | Preview

Abstract

AIoT is the combination of Internet of Things (IoT) and Artificial Intelligence (AI) technologies. While IoT emphasizes more on scalable and efficient communications, AI focuses more on reproducing human capabilities such as recognition and forecasting. An efficient AIoT platform may not be obtained directly from integrating existing IoT and AI serving platforms by considering the AIoT service reproduction and evolution. In this work, we propose an AIoT platform that empowers developers to build sophisticated and scalable applications. Our platform is derived based on exchange-based RabbitMQ broker and Advanced Message Queuing Protocol (AMQP) to facilitate the communications among heterogeneous data sources and AI models. By incorporating an AMQP broker, it supports diverse data exchanges, AI models chaining, and flexible message routing and processing. AI models can be deployed efficiently through containerization with flexible and shared data paths to facilitate computations. Hence, developers can focus on service and application requirements. We also present a case study in smart healthcare to validate our design.

Item Type: Book Section
Uncontrolled Keywords: sdg 3 - good health and well-being ,/dk/atira/pure/sustainabledevelopmentgoals/good_health_and_well_being
Faculty \ School: Faculty of Science > School of Computing Sciences
Related URLs:
Depositing User: LivePure Connector
Date Deposited: 13 Nov 2023 18:10
Last Modified: 14 Nov 2023 11:23
URI: https://ueaeprints.uea.ac.uk/id/eprint/93632
DOI: 10.1145/3616391.3622770

Actions (login required)

View Item View Item