A global optimization heuristic for the decomposed static anticipatory network traffic control problem anticipatory network traffic control problem

Developing traffic control strategies taking explicitly into account the route choice behavior of users has been widely recognized
irregularities in the solution space shape, such as non-convexity and non-smoothness. In this work, we propose an extended
as a very challenging problem. Furthermore, the inclusion of user behavior in optimization based control schemes introduces strong
decomposition scheme for the anticipatory traffic control problem, based upon our previous contributions, which aims at i) reducing
irregularities in the solution space shape, such as non-convexity and non-smoothness. In this work, we propose an extended
the computational complexity of the problem by approaching it in a controller-by-controller fashion, and ii) internalizing specific
decomposition scheme for the anticipatory traffic control problem, based upon our previous contributions, which aims at i) reducing
constraints in the objective function, guiding the optimization process away from non-significant minima, such as flat regions.
the computational complexity of the problem by approaching it in a controller-by-controller fashion, and ii) internalizing specific
Through two small scale test networks and different, randomly chosen initial points, we compare how the proposed extension
constraints in the objective function, guiding the optimization process away from non-significant minima, such as flat regions.
influences optimization results with respect to our previously developed decomposed approach, as well as centralized schemes.
Through two small scale test networks and different, randomly chosen initial points, we compare how the proposed extension influences optimization results with respect to our previously developed decomposed approach, as well as centralized schemes.