This paper develops a traffic simulator for urban mobile network of Manhattan type. This simulator permitted to measure the performance of a network by varying its parameters: Many interesting conclusions resulted from the simulations realized. For instance, when some channels of each cell are reserved for handoffs, a consequent decrease in the loss probability for current calls has been observed. However, in order to not excessively penalize new calls, the number of reserved channels should not exceed the half of the total number of channels. On the other hand, the remarkable aspect evidenced by these simulations is the significant improvement of the network performance, thereby a greater quality of service supplied to subscribers, when frequency assignment of dynamic type is used. Therefore, the improvements of network performance are not proportional to the number of cells consulted during the frequency assignment process.
