Traffic engineering cooperating with traffic monitoring for the case with incomplete information

Traffic Engineering (TE) accommodates traffic efficiently by dynamically configuring the routes so as to follow traffic changes. If traffic changes frequently and drastically, the interval to perform the route reconfiguration should be set to a short value, to follow traffic changes. To shorten the interval, obtaining the traffic information becomes a problem; measuring traffic information accurately of the whole network, which is required to calculate the suitable route, is difficult to be obtained in a short interval, due to the overhead to monitor and collect the traffic information. We have proposed the framework of the TE for the case that only a part of traffic information can be obtained at each time slot. This framework was inspired by the human brain mechanism. In this framework the conditional probability is considered to make decisions. In this framework, a controller is deployed. The controller (1) obtains a limited number of traffic information, (2) estimates and predicts the probability distribution of the traffic, (3) configures the routes considering the probability distribution of the predicted future traffic, and (4) selects the traffic to be monitored at the next period considering the performance of the route reconfiguration using the traffic information obtained at the next period. In this paper, we discuss the details of the each step of our framework. Then, we evaluate our framework.