Stochastic user equilibrium assignment with traffic-responsive signal control

This paper considers the Stochastic User Equilibrium (SUE) assignment problem for a signalcontrolled network in which intersection control is flow-responsive. The problem is addressed within a Combined Traffic Assignment and Control (CTAC) modeling framework, in which the calculation of user equilibrium link flows is integrated with the calculation of consistent signal settings. It is assumed that network equilibrium is dispersed due to user misperceptions of travel times, and that the intersection control system is designed to allow the persistent adjustment of signal settings in response to traffic flow variations. Thus, the model simulates real-world situations in which network users have limited information and signal control is traffic-actuated. The SUE-based CTAC model is solved algorithmically by means of the socalled Iterative Optimization and Assignment (IOA) procedure, a widely used heuristic which relies on the alternate execution of a control step (signal setting calculation for fixed link flows) and an assignment step (network equilibration under fixed signal settings). The main objective of the study is to define a methodological framework for the evaluation of the performance of various traffic-responsive signal control strategies in interaction with different levels of user information, as represented by the spread parameter of the perceived travel time distribution assumed in the SUE assignment submodel. The results are of practical relevance in a policy context, as they provide a basis for assessing the potential integration of Advanced Traveler Information Systems (ATIS) and signal control systems. Several computational experiments are carried out on a small, contrived network and using realistic intersection delay functions, in order to test the behavior of the model under a wide range of conditions; in particular, convergence pattern and network performance measures at equilibrium are analyzed under alternative information/control scenarios and for various demand levels. The issue of uniqueness of the model solution is addressed as well.