Multi-time Step Traffic Queue Prediction Model based on Discrete Time Point Process (DTPP)

An accurate prediction of queue length at the intersection has been one of the key issues in implementation of the adaptive signal control systems as well as the bus signal priority systems (BSP). Because of the analytical queueing models' nature, most prediction models have limited in predicting the queue length for longer time interval. This paper develops a multi-time step queue prediction model based on DTPP to forecast multi time steps using the delay between the time vehicle is detected and the time vehicle arrived at the end of queue. The test on simple one-way network provided a fairly accurate result in queue prediction, less than I vehicle in average. For practical application to the traffic signal system in operation, it is required to improve further the model being able to consider left-turn and right-turn traffic at the intersection.