HUO, JINBIAO (Southeast University), SHI, XINYU (Southeast University), Gu, Ziyuan (Southeast University), Liu, Zhiyuan (Southeast University)

Calibration of Traffic Simulation under Heteroscedastic Noise: A Bayesian Optimization Approach

Scheduled for presentation during the Invited Session "S19b-Artificial Transportation Systems and Simulation" (TH-EA-T19), Thursday, November 20, 2025, 14:50−15:30, Surfers Paradise 1

2025 IEEE 28th International Conference on Intelligent Transportation Systems (ITSC), November 18-21, 2025, Gold Coast, Australia

This information is tentative and subject to change. Compiled on October 18, 2025

Abstract

This study investigates the parameter calibration problem in traffic simulation. While extensive research has yielded effective solutions, the issue of heteroscedastic simulation noise—a common feature in traffic models—remains largely unaddressed. Such noise introduces varying levels of uncertainty across the solution space, making efficient use of limited computational resources particularly challenging. Existing calibration methods often adopt uniform simulation budgets, overlooking noise variability and leading to suboptimal solutions. To address this, we propose a two-stage calibration framework based on Bayesian optimization. The first stage employs classical BO to explore promising regions of the parameter space, while the second stage reallocates the remaining simulation budget to improve the evaluation of selected solutions. This design balances exploration and exploitation under budget constraints. Experiments on real-world traffic data demonstrate that the proposed method improves simulation efficiency and robustness in the presence of heteroscedastic noise.