Embedded-Based Battery Management System using TinyML and IoT: A Review

Abstract

Lithium-ion batteries require management systems that combine dependable protection with increasingly accurate state estimation and predictive maintenance. This review critically examines embedded battery management systems that integrate Tiny Machine Learning (TinyML) and Internet of Things (IoT) connectivity. Nineteen unique studies published from 2022 to 2026 were synthesized across four dimensions: monitored variables and predicted states, learning approach, embedded deployment evidence, and communication architecture. The literature shows strong progress in data-driven estimation of state of charge, state of health, remaining useful life, temperature, and abnormal cell behavior. However, the evidence is fragmented: many studies validate algorithms offline, a smaller group reports true microcontroller deployment, and only a limited subset closes the loop from prediction to safety action while maintaining secure remote monitoring. The review therefore proposes a layered reference architecture in which deterministic protection remains local, TinyML performs low-latency inference at the edge, and IoT services support supervisory visualization, fleet analytics, and controlled model updates. Key research priorities are cross-chemistry validation, standardized edge benchmarks, uncertainty-aware inference, cybersecurity, drift monitoring, and hardware-in-the-loop evaluation. The synthesis provides a practical roadmap for low-cost, energy-efficient, and trustworthy intelligent BMS development.