The cordon-based Electronic Road Pricing (ERP) system in Singapore adopts the average travel speed as an index for evaluating the traffic congestion within a cordon area, and the maintenance of the average travel speed within a satisfactory range is taken as the objective of the toll adjustment. To formulate this practical speed-based toll design problem, this paper proposes a mathematical programming with equilibrium constraint (MPEC) model with the objective of maintaining the traffic condition in the cordon area. In the model, the network users' route choice behavior is assumed to follow probit-based stochastic user equilibrium with elastic demand, asymmetric link travel time functions and continuous value-of-time. A distributed revised genetic algorithm is designed for solving the MPEC model. Finally, a network example based on the ERP system is adopted to numerically validate the proposed models and algorithms, and further indicates that the computation speed can be improved greatly by using a distributed computing system.