Improving the accuracy and efficiency of online calibration for simulation-based Dynamic Traffic Assignment

Simulation-based Dynamic Traffic Assignment models have important applications in real-time traffic management and control. The efficacy of these systems rests on the ability to generate accurate estimates predictions states, which necessitates online calibration. A widely used solution approach for calibration is Extended Kalman Filter (EKF), -- although appealing its flexibility incorporate any class parameters measurements poses several challenges with regard accuracy scalability, especially congested situations large-scale networks. This paper addresses issues turn so as improve efficiency EKF-based approaches large First, concept state augmentation revisited handle violations Markovian assumption typically implicit EKF. Second, a method based graph-coloring proposed operationalize partitioned finite-difference that enhances scalability gradient computations. Several synthetic experiments real world case study demonstrate application yields improvements terms both prediction computational performance. work has real-world deployments simulation-based dynamic assignment systems.