OPTIMISATION OF TRAFFIC SIGNALS USING A CELL TRANSMISSION MODEL

Management of transportation networks especially in major cities is impossible without traffic signals, and it is widely accepted that many practical problems remain to be solved. Evolution of traffic flows over complex networks can be predicted over time using macroscopic models, which deal with traffic flow as if it were a fluid, or microscopic ones, which, on the other hand, follow the movement of individual vehicles through time. For example, in microscopic simulation models of traffic such as Paramics or Vissim each vehicle is separately identified and tracked and has its own characteristics, for instance, desired speed, desired separation, reaction time, destination, vehicle type and so on, and produce a lot more output than is ever normally needed. Another problem connected with microscopic models is that because of their complexity the run time for a realistic network can take much more time than it takes in reality, and optimisation of signal settings is difficult because of the need for many repeat runs and “noisy” output, due to the Monte Carlo nature of such models.