With the recent spread of mobile devices like smart phones and tablets, the proportion of mobile traffic in the total Internet traffic has been continuously increasing. Particularly, the utility of users, which is a quality metric of user experience in the service, decreases when a lot of mobile traffic concentrates at a wireless access network at a specific time. Therefore, in this study, we
... [Show full abstract] instruct users to delay their traffic to move a part of traffic in the peak time to the off-peak time and to smooth traffic temporally. However, since users do not always follow the instructing message from the communication system, we consider the response of users according to their utility in this study. We also consider two types of user models: learning and non-learning. The former considers past experienced utility, while the latter only considers the current utility without considering their past experience. In this paper, we assume both models and will verify our system works well.