Mobile network operators face congestion problem which degrades the Quality of Service. Previous studies on telecommunication service pricing show that it is an efficient mean to manage network profitability and traffic regulation. The aim of this work is to propose pricing policy which is able to reduce congestion and increase profitability in mobile networks. In this paper, we investigate several pricing models in mobile networks, and we propose a new model which consists in periodically adjusting the communications prices. The model applies different taxes per period in order to reduce the network load during congested periods and increase the load during off-peak ones. We use probabilistic approach based on utility function to predict user's behaviour that allow us to predict the offered traffic and manage it according to the network capacity.