QoS Model of a Router with Feedback Control

The Internet has evolved into a shared, integrated platform of a broad range of applications with different Quality-of-Service (QoS) requirements. Routers are an important part of the Internet and play a critical role in assuring QoS. A router is usually placed between two networks to receive data packets from one network and then transmit those data packets to another network if necessary. Data packets are the actual units of data traveling on computer networks. A data packet has two parts: header and data. The data carries messages, such as e-mail text, from computer applications. The header carries information that is required to control and manage the transmission of the data packet on computer networks. Existing approaches for providing QoS involve prediction or estimation for traffic characterization to determine parameters required of static traffic admission control. However, prediction or estimation inaccuracy in traffic characterization can result in inappropriate parameter settings for static admission control and, in turn, compromise QoS or resource utilization. This study presents a QoS model of a router with feedback control that monitors the state of resource usage and adaptively adjusts parameters of traffic admission control to overcome prediction or estimation inaccuracy and achieve a balance between QoS and resource utilization. The QoS model of a router with feedback control is simulated to test its performance on QoS and resource utilization in both heavy and light traffic conditions. The performance of the QoS model of a router with feedback control is also compared with that of two basic QoS models of a router with static admission control using admission control parameters resulting from over- and under-characterization of traffic, respectively. The simulation results show that the QoS model of a router with feedback control achieves a better balance between QoS and resource utilization than the basic QoS models with over- and under-characterizations of traffic in the heavy traffic condition. This study also shows that the three models of routers demonstrate similar QoS performances and resource utilization in the light traffic condition. Copyright © 2005 John Wiley & Sons, Ltd.