A Traffic Predictor Based on Adaptive Fuzzy Clustering Technique for IP-based Networks

Abstract

A traffic predictor is significant to ensure the reliability and utilization of network resource in the next generation Internet. An adaptive approach based on fuzzy clustering technique for traffic estimation in resource protocols is proposed. The proposed traffic estimator is deployed to learn the knowledge about the traffic flows and generates a set of cluster representing the profile of flow characteristics. Different from the past k-mean clustering, the proposed clustering algorithm does not need the prior information of cluster number. The size of cluster set is fully determined by target pattern set. By matching the derived cluster set of flow patterns with the current flow pattern, the future flow is estimated and predicted. In experiments, two kinds of traffic flow are used to verify the algorithm performance, periodic flow and Poisson flow. Results show excellent performance of the proposed predictor. The prediction errors, in average, are near 15.4% and 3.56% for periodic flow and Poisson flow, respectively.