Digital Twin Based Open Platform for IoT Offloading Control: Enabling System Transparency and User Participation

Abstract

In the integrated infrastructure of Society 5.0, efficiently managing complex IoT (Internet of Things) systems is crucial. IoT devices frequently encounter limitations like limited network bandwidth, storage, and processing power, which lead to data congestion and reliance on cloud computing. Recent advancements in edge computing have emphasized hybrid offloading methods. Despite these technological strides, managing IoT systems still presents significant challenges. Typically, these systems are complex and closed, making it difficult for non-specialists to engage with and manage them effectively. This complexity is contrary to the principles of open platforms, which aim to democratize access and control, allowing a broader range of stakeholders to participate in system management. In this paper, we propose a novel system that integrates IoT offloading with a digital twin (DT) on an open web platform, enhancing both the transparency and the inclusivity of IoT system management. The system utilizes a machine learning model to dynamically select the most efficient data processing routes-either edge or cloud-based on real-time network conditions and data volume. Our case study, focused on person detection, employed linear regression and decision tree regression to optimize offloading decisions, achieving accuracies of 73.9 % and 95.7 % respectively. This setup not only demonstrated the effectiveness of the proposed system in real-world scenarios but also highlighted its capability to provide seamless integration and control through a DT interface, thus making complex IoT systems more accessible and manageable.