Research On QoS-Based DiffServ Model

Traditional Internet doesn't provide Quality of Service guarantees. The network can only provide best effort data delivery service. The best effort service is generally enough for traditional applications like Web, FTP and E-mail etc., because these applications do not have real-time requirements. With the advent of high-end applications and the development of grid techniques, more and more network applications are emerging such as Voice over IP, Video Conference, Distance Education and Data Grid etc. which have stricter request. These high-end applications differ greatly from traditional applications in that they are real time, mutual, abrupt, dynamic and delivery data are gigantic. The best effort service is no longer adequate for these applications. How to provide Quality of Service guarantees to high-end network applications is a hot topic in current network research.DiffServ model is one important model of the IP QoS which was put forward by IETF only several years ago. This model works on the basis of packet classification granularity, it can overcome the limitation of the IntServ/RSVP model and possesses good expansibility and robustness, so it is easy to be realized on the large-scale network and could be the framework of the future Internet(IPng). This paper, on the basis of the DiffServ model architecture, studies the correlative contents of this mode, for example, queue management and scheduling mechanism.This paper consists of five chapters. The first chapter analyzes the high-end network applications on the Internet with rapid development and the QoS requirements of them, discusses the definition of the QoS and investigates study status of the DiffServ model at domestic and international. The second chapter introduces correlative basic contents of the DiffServ model. The next chapter studies queue management of the DiffServ model. The fourth chapter studies queue schedule of the DiffServ model and gives schedule scheme. The fifth chapter studies integration of IntServ model and DiffServ model. In the end, the thesis is summarized and the prospect of the future research in DiffServ model is pointed out.