An AR model based on a fuzzy approach for traffic modeling

Broadband integrated digital networks (B-ISL)N) using an asynchmnous transfer mode (ATM) is the new generation of ISDNs for telccommunicalion networks. An ATM is designed to support various classes of multimedia traffic flow (voice, data, and video) with diffcrcnt bit rate and quality of senice (QoS) requiremenb. However, in actual networks, congcstion and deadlock may occur due lo some problem. An AMT provides various trafic controls to rcducc cungstion and to keep high quality networking such as conncction admiuion control (CAC), usage parametcr control (UPC), priorily control (PC). Thcse controls are essenlially reactive-type control. For proaelive control, future cell rate flowing into thc ATM has to be predicted on the basis of the past and current traffic, and control signal then generatcd using the prcdicted cell rale. We propose a traffic model by an autoregressive (AR) model based on fuzzy approach in this paper. Compulational experiments were carried out to show uscfulness of lhe proposed model. As a result of experiments, the proposed model shows better performance than that by previous models.