Online calibration with SUMO for network-wide traffic and emission monitoring – Case study ITS Huainan

Flötteröd, Yun-Pang; Behrisch, Michael · 2022 · Crossref

DOI: 10.52825/scp.v2i.88

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper presents the implementation of an online calibration and prediction system using the microscopic traffic simulation software SUMO, integrated into the KeepMoving traffic information platform for the Intelligent Transportation System (ITS) of Huainan, China. The research addresses the challenge of monitoring network-wide traffic states and emissions in real-time, a task difficult to achieve with sparse stationary sensors alone. By combining real-time sensor data with simulation, the system provides decision support for traffic management, offering both current traffic conditions and short-term predictions, particularly in areas lacking physical detectors. The methodology involves a Python-based process chain that handles data processing, simulation, and calibration. Data from 709 stationary sensors (microwave and camera-based) and floating car data (FCD) undergo correction, aggregation, fusion, and extrapolation. The simulation network, derived from OpenStreetMap and refined with satellite imagery, comprises approximately 10,000 edges and 5,000 nodes. Traffic demand is based on a 2009 origin-destination matrix disaggregated into 81 traffic analysis zones, yielding an estimated 1.34 million daily trips. The system operates in 5-minute intervals using mesoscopic simulation settings. Calibration adjusts simulated speeds by modifying maximal speeds and adjusts flows by inserting or removing vehicles from route sets to match measured data. If data is unavailable, the simulation continues without calibration. Preliminary results from a two-week testing period in January 2021 demonstrate promising calibration performance. At the 5-minute interval level, approximately 80% of calibrated flows and over 95% of calibrated speeds exhibited an absolute relative error of 15% or less. At the hourly level, 92% of calibrated flows achieved a GEH statistic value of 5 or lower, indicating a good fit between simulated and real data. Case studies of specific edges showed that the system effectively tracked speed fluctuations and flow changes, although discrepancies occurred during congestion when vehicle insertion failed due to lack of space on edges. The system maintained a total processing time of less than 5 minutes per interval, meeting real-time operational requirements. The study concludes that online calibration with SUMO is a viable approach for network-wide traffic and emission monitoring, providing valuable data for traffic management despite limitations in sensor coverage and data gaps. The authors note that the simulation’s defensive calibration strategy, which delays vehicle insertion to avoid conflicts, can occasionally hinder flow accuracy during congestion. Future improvements are expected through the expansion of sensor networks, updates to the traffic demand model to reflect ongoing urban development, and the integration of additional data sources such as mobile phone and social media data.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• in vehicle coaching