Revisiting Level of Service Assessment by Incorporating Speed Variability and Acceleration Noise Using High-Resolution Trajectory Data

Abstract

Level of service (LOS) is a fundamental performance measure in traffic engineering practice, widely used to evaluate roadway operating conditions and based primarily on average traffic measures such as mean speed. While effective for standardization, conventional LOS assessment may fail to capture the dynamic and stochastic nature of traffic flow, particularly under congested and transitional conditions. This study revisits LOS assessment by explicitly incorporating traffic flow variability into the evaluation framework. Traffic is conceptualized as a stochastic dynamic system in which traffic states with similar mean speeds may exhibit substantially different stability characteristics. Building on this premise, a variability-based LOS framework (LOS-V) is proposed, integrating standardized speed variability and acceleration noise as indicators of traffic stability. The proposed framework is evaluated using high-resolution vehicle trajectory data from the Next Generation Simulation dataset. A cross-classification analysis between conventional LOS and LOS-V reveals substantial redistribution across service categories, particularly for intermediate traffic states. Sensitivity analysis further confirms that the LOS-V classification is robust to variations in weighting coefficients. The results demonstrate that average speed alone is insufficient to distinguish between stable and unstable traffic states and may mask important operational deficiencies. Overall, the LOS-V framework enhances the interpretability and diagnostic capability of LOS assessment by providing a more behaviorally meaningful representation of traffic operating conditions in data-rich traffic environments.