Enhancing Edge Analytics Using TinyML and IoT: Toward Energy-Efficient and Intelligent Data Processing in Distributed Environments | ||
| Journal of Communication Sciences and Information Technology | ||
| Volume 2025, Issue 3, October 2025, Pages 25-38 PDF (1.17 M) | ||
| Document Type: Original Article | ||
| DOI: 10.21608/jcsit.2025.438575.1022 | ||
| Author | ||
| Mariam Mahmoud Hagag* | ||
| Department of Data science, faculty of artificial intelligence, Egyptian Russian University, badr, Cairo, Egypt | ||
| Abstract | ||
| The convergence of Internet of Things (IoT) ecosystems with Tiny Machine Learning (TinyML) has redefined the paradigms of distributed analytics and computational intelligence. Traditional cloud-centric models impose high latency, excessive energy consumption, and increased privacy risks, limiting real-time responsiveness for mission-critical applications. This paper introduces an integrated edge analytics framework leveraging TinyML-enabled IoT devices for low-latency, energy-efficient, and adaptive decision-making. The proposed architecture unifies lightweight embedded learning, confidence-triggered inference offloading, and collaborative edge-to-cloud synchronization to enhance analytic performance at the periphery of the network. Empirical simulations conducted on embedded-class processors demonstrate latency reduction exceeding 50% and energy savings up to 42% compared to conventional edge–cloud paradigms, without compromising inference accuracy. This study contributes a scalable foundation for intelligent IoT infrastructures capable of performing dynamic analytics under constrained computational and communication environments, advancing the vision of sustainable and autonomous edge intelligence. Keywords: TinyML, Edge Analytics, IoT, Embedded Intelligence, Energy Efficiency, Adaptive Offloading, Distributed Learning. | ||
| Keywords | ||
| TinyML; Edge Analytics; IoT Embedded Intelligence; Energy Efficiency, Adaptive Offloading | ||
|
Statistics Article View: 16 PDF Download: 4 |
||
