Perspectives of TinyML-Based Self-management in IoT-Based Systems

The emergence of the Internet of Things (IoT) has brought with it the need to efficiently and autonomously manage the systems that power these networks of connected objects. With this in mind, the use of TinyML (Tiny Machine Learning) presents itself as a promising technique to the self-management of IoT-based systems. TinyML enables the execution of Machine Learning (ML) models directly on IoT devices with low computing and memory power, offering benefits such as reduced bandwidth requirements, real-time responsiveness, data privacy preservation and resource optimization. Through proactive monitoring, anomaly detection, performance problem diagnosis and real-time adaptation, TinyML can facilitate the efficient self-management of IoT-based systems, paving the way for significant improvements in performance, efficiency, security and sustainability of these systems. With this in mind, we explore the possibilities offered by TinyML for the self-management of IoT-based systems, examining its advantages, limitations and future opportunities.