A collaborative cloud-edge adaptive dual-graph model achieved improved passenger flow forecasting performance compared with baseline methods on a real-world, large-scale rail transit dataset.
Accurate passenger flow predictions support station operations, resource allocation, and service optimization across rapidly expanding high-speed railway networks. However, passenger volumes fluctuate over time, across station spatial interactions, and through regional travel dependencies.
Existing forecasting methods that rely on fixed spatial structures or centralized computing architectures face challenges in capturing dynamic passenger correlations while meeting the real-time and large-scale demands of railway service applications.
To address these constraints, the proposed framework employs a dual-graph representation that pairs physical station connectivity with an input-window-adaptive correlation graph derived from recent passenger flow observations.
A dedicated spatio-temporal learning module captures multi-scale temporal variations alongside complex spatial dependencies across the rail transit network.
The collaborative system integrates global knowledge from the cloud side with local real-time features on the edge side, maintaining a lightweight edge-side inference branch while validating the architectural components in smart railway service scenarios.
The research, conducted by Ling Liu and Majid Ghani Varzaneh, appeared in the Journal of Cloud Computing on September 11, 2026. Published as an open-access, peer-reviewed accepted article carrying a permanent DOI, the work provides early access to the findings ahead of the final Version of Record.
In their experimental evaluation, the researchers tested the architecture on a real-world, large-scale rail transit passenger flow dataset. The empirical results demonstrated that the collaborative forecasting model achieved better forecasting performance than the compared baselines. Subsequent empirical analyses further verified the specific effectiveness of each component, demonstrating the practical advantages of the cloud-edge design for real-time railway service scenarios.
Subject of Research: High-speed railway passenger flow forecasting
Article Title: Cloud-Edge collaborative adaptive dual-graph spatio-temporal passenger flow forecasting model for smart high-speed railway services
Article References: Liu, L., & Varzaneh, M. G. (2026). Cloud-Edge collaborative adaptive dual-graph spatio-temporal passenger flow forecasting model for smart high-speed railway services. Journal of Cloud Computing. https://doi.org/10.1186/s13677-026-00986-3
Image Credits: AI Generated
DOI: 10.1186/s13677-026-00986-3
Keywords: passenger flow forecasting, high-speed railway, cloud-edge collaboration, dual-graph model, spatio-temporal learning, rail transit, station operations, resource allocation, service optimization, physical connectivity, adaptive correlation graph, edge computing, smart railway services, temporal fluctuations, spatial dependencies, lightweight edge inference, transit network modeling
News Source: Denise Maddox. (November 19, 2026). Adaptive Cloud-Edge Dual-Graph Model Improves High-Speed Railway Passenger Flow Forecasting. Scienmag.



