Intelligent task offloading for sustainable energy management in industrial IoT edge cloud systems

Efficient task offloading is critical for the Industrial Internet of Things (IIoT), yet existing strategies often fail to address the complex inter-task dependencies inherent in real-world industrial workflows. Current approaches predominantly treat tasks as independent units, an oversight that leads to execution deadlocks, suboptimal resource scheduling, and system instability. To bridge this gap, this study introduces M-SIEGFENNet-TUA, a Multi-Scale Integrated Edge-Guided Finite Element Neural Network with Tactical Unit Algorithm, designed for sustainable and reliable task offloading in edge–cloud environments. The model integrates multi-scale analysis for feature extraction, edge-guided attention for prioritization, and finite element optimization for balanced resource allocation. Crucially, unlike standard models, M-SIEGFENNet-TUA explicitly manages task dependencies, ensuring that interconnected tasks are executed in an optimal sequence. Experimental results demonstrate reduced latency (120–160 ms), lower delay (cloud 50→7 ms), high task completion rates (up to 94%), and minimal energy consumption (52.7 J). Overall, the framework improves efficiency, reliability, and sustainability, aligning with SDG 7 (Affordable and Clean Energy) and SDG 11 (Sustainable Cities and Communities).