Dynamic Edge Orchestration for Low-Latency IoT Networks: A Comparative Deep Learning Framework for Network Performance and Reliability

Abstract

The rapid growth of the Internet of Things (IoT) has increased the demand for intelligent edge computing solutions capable of delivering low-latency and reliable services. However, most existing edge architectures remain conceptual and lack data-driven mechanisms for latency prediction and adaptive task orchestration. This paper proposes the Dynamic Edge Orchestration Platform (DEOP), an AI-enabled edge computing architecture that integrates IoT devices, edge gateways, edge nodes, a Dynamic Orchestration Engine (DOE), and cloud infrastructure to support latency-aware resource management in Multi-access Edge Computing (MEC) environments. To enable intelligent orchestration, three machine learning models Decision Tree Regression, Support Vector Regression, and a one-dimensional Convolutional Neural Network (1D-CNN) are developed and evaluated using the Edge-IIoTset dataset. The proposed framework bridges the gap between conceptual edge architectures and intelligent decision-making by combining predictive analytics with dynamic orchestration, providing a practical and scalable foundation for next-generation edge-enabled IoT applications.