Edge MLOps framework for AIoT applications

Abstract

Recent years witnessed a boom in IoT devices resulting in big data and demand for low latency communication giving rise to a demand for 5G Networks. This shift in the infrastructure is enabling real-time decision making using artificial intelligence for IoT applications. Artificial Intelligence of Things (AIoT) is the combination of artificial intelligence (AI) technologies with the Internet of Things (IoT) infrastructure to achieve more efficient IoT operations and decision making. Edge computing is emerging to enable AIoT applications. Edge computing enables generating insights and making decisions at the data source, reducing the amount of data sent to the cloud and central repository. An ecosystem to facilitate edge computing for AIoT applications has become essential to make real-time decisions at the data source. In this thesis, we develop a framework to facilitate machine learning at the edge for AIoT applications which enables continuous delivery, deployment and monitoring of Machine Learning models at the edge (Edge MLOps). We will propose an ideal architecture, services, tools and methods for optimization of costs, operations, and resources to facilitate efficient edge-cloud operations at scale using Microsoft Azure. Validation of the framework is done by performing iterative experiments with IoT devices set up in rooms on a campus enabled by a private LAN, this campus is based in Helsinki, Finland. For experiments, multivariate time series forecasting is done to predict future air quality in respective rooms using machine learning models deployed in respective edge devices. We conclude these AIoT experiments to validate proposed edge MLOps framework efficiency, robustness, scalability and resource optimization.