Traffic estimation in high-speed communication networks using fuzzy systems

We present a practical application of fuzzy systems in communication networks. In this case a fuzzy relational model is developed to estimate the bytes transferred every time unit over a communication link. Such estimation is useful to provide an algorithm to dynamically reserve resources in order to guaranty the quality of service of the communications. In order to evaluate the fuzzy estimator we apply the information about the whole IP packets received at the router of the Public University of Navarra along two months. The estimator predicts the load of the communication link 1, 2, 4 and 6 hours before. The results show that the estimator can be used to dynamically assign bandwidth for communication services.