Article: Edge IoT causal graph neural network for enhancing urban carbon sink accounting accuracy Journal: International Journal of Information and Communication Technology (IJICT) 2025 Vol.26 No.51 pp.85 - 107 Abstract: Global climate change has made accurate urban carbon sink accounting crucial for low-carbon policies and ecological restoration, yet traditional methods suffer from inefficiency, low precision, and poor generalisation. To address these issues, this study proposes an edge IoT-causal graph neural network framework. It integrates a multi-layer edge internet of things architecture reducing data transmission latency by over 40% compared to cloud-centric systems. Additionally, a causal graph neural network model is developed; it infers the causal structure of environmental variables via an improved PC algorithm and embeds this structure into graph attention network training to avoid spurious correlations. Experimental validation on real urban green space data shows the framework achieves 94.7% accounting accuracy, outperforming traditional graph neural networks, support vector machines, and remote sensing inversion by over 8.5%. This work provides a practical technical paradigm for high-precision urban carbon sink accounting, supporting evidence-based urban low-carbon management. Inderscience Publishers - linking academia, business and industry through research

Title: Edge IoT causal graph neural network for enhancing urban carbon sink accounting accuracy

Authors: Zhenyu Qian; Qingfeng Zhang; Xiaohong Liu; Youshui Zhang; Zixin Jia

Addresses: College of Natural Resources and Environment, Northwest A&F University, Yangling, 712100, China ' College of Natural Resources and Environment, Northwest A&F University, Yangling, 712100, China ' Shaanxi Forestry Survey and Planning Institute, Xi'an, 712100, China ' Shaanxi Forestry Survey and Planning Institute, Xi'an, 712100, China ' College of Natural Resources and Environment, Northwest A&F University, Yangling, 712100, China

Abstract: Global climate change has made accurate urban carbon sink accounting crucial for low-carbon policies and ecological restoration, yet traditional methods suffer from inefficiency, low precision, and poor generalisation. To address these issues, this study proposes an edge IoT-causal graph neural network framework. It integrates a multi-layer edge internet of things architecture reducing data transmission latency by over 40% compared to cloud-centric systems. Additionally, a causal graph neural network model is developed; it infers the causal structure of environmental variables via an improved PC algorithm and embeds this structure into graph attention network training to avoid spurious correlations. Experimental validation on real urban green space data shows the framework achieves 94.7% accounting accuracy, outperforming traditional graph neural networks, support vector machines, and remote sensing inversion by over 8.5%. This work provides a practical technical paradigm for high-precision urban carbon sink accounting, supporting evidence-based urban low-carbon management.

Keywords: edge IoT; causal graph neural network; CGNN; urban carbon sink accounting; causal inference; multi-source data fusion.

DOI: 10.1504/IJICT.2025.151075

International Journal of Information and Communication Technology, 2025 Vol.26 No.51, pp.85 - 107

Received: 19 Sep 2025
Accepted: 19 Oct 2025
Published online: 12 Jan 2026 *

Title: Edge IoT causal graph neural network for enhancing urban carbon sink accounting accuracy

Keep up-to-date